Diathesis Oppositions and Verb Morphology. Present and Aorist Systems in Ancient and Modern Greek

1.1. This paper deals with Ancient Greek verb morphosyntax and has two different, but complementary, purposes. The first is to give a detailed description of the relationship between forms and functions in the aorist and present systems of Ancient Greek. The second goal is to propose an analysis of the diachronic changes affecting relevant aspects of verb morphosyntax from Ancient to Modern Greek.

In Ancient Greek both the aorist and the present systems feature two inflectional paradigms that contrast with respect to the syntactic properties of the Subject. On the other hand, in Modern Greek no inflectional contrast is found in the aorist system, but the present system is similar, in this respect, to that of Ancient Greek. This is an important difference between the two stages of the language that may be accounted for in terms of markedness relationships between forms and functions.

For this research, I have taken into account all the constructions in which finite verb forms in the aorist and present occur. They have been collected from the following literary works: Homer’s poems and hymns, Hesiod’s poems, and Herodotus’ Histories (see Appendices (1)-(5) at the end of this paper for an illustration of all the contructions that occur in my corpus: I give only one example for each lexical item).¹

‘for the most part of their men were slain by the Persians’

(1b) autoús t’ékteinon […] (Od. XIV 265)

they-acc.pl and slay-ind.impf.3pl.act

‘and [they] killed the men’

‘he who was your comrade is slain’

(2b) hēmeîs d’hérma pólēos apéktamen […] (Od. XXIII 121)

we-nom.pl so stay-acc.sg city-gen.sg kill-ind.aor.1pl.act

‘whereas we have been killing the stay of a whole city’

Contrary to what we find in the present system, however, the relationship between the passive vs. active contrast and inflectional contrast is not productive in the aorist system. In fact, this relationship is restricted to the morphological classes of thematic aorist (e.g. éskheto ‘was stayed’ and dieprátheto ‘was sacked’) and root aorist (e.g. éblēto ‘has been struck’). From a lexical point of view, these classes are not productive, compared to the so-called sigmatic aorist, whose morpho-lexical productivity is well-known (cf. Willi 2010: 512). This fact is interesting when seen in the light of another property frequently emphasized by scholars: the sigmatic forms do not occur in passive structures (cf. Wackernagel 1926: 137-8, Prévot 1935: 17-8, Schwyzer 1959: 757 and Ambrosini 1996: 21).

‘he was laid in the burial-place built for himself in the temple’

(3b) […] ho dé min éthapse en têi boï̀ taútēi (Hdt. II 131)

he-nom so she-acc bury-ind.aor.3sg.act in det-dat.sg cow-dat.sg this-dat.sg

‘then he buried her in this image of a cow’

‘even so were the Achaeans utterly panic-stricken by Hektor and father Zeus’

(4b) […] toùs d’állous Danaoùs ephóbēse Kroníōn (Il. XI 406)

det-acc.pl so other-acc.pl Danaan-acc.pl strike-ind.aor.3sg.act son of Kronos-nom.sg

‘for the son of Kronos has struck the rest of the Danaans with panic’
Examples (3b) and (4b) include verb forms that belong to the sigmatic class. From a morphological point of view, the sigmatic affix -s(a)-⁵ combines with inflectional endings. The relationship between inflection and syntax is trivial: there is active inflection in active constructions, as already observed in examples (1b) and (2b). Examples (3a) and (4a), on the other hand, involve the so-called “passive aorist”. As far as morphology is concerned, these forms (e.g. etáphē and ephóbēthen) are characterized by the combination of the affixes -ē- or -thē- and inflectional endings. From a morpho-syntactic point of view, pairs of clauses such as those in (3)-(4) are different from each other, the first one being passive and the second one active, but they do not contrast with each other by means of verb inflection, as was the case in (1)-(2). Instead, in all the clauses in (3)-(4), verb inflection is invariably active.
1.4. In order to account for this phenomenon, many scholars assume an ontological perspective. They suggest we should study the category of “passive aorist” by itself, without considering that these forms, far from being isolated, belong to a larger verb system and therefore their functional value should be determined with reference to that system as a whole. Comparative linguists often assign a semantic value like “stative” or “fientive” to one of the “passive aorist” affixes, i.e. -ē-, and they suggest that the passive use of these forms developed from the “stative” or “fientive” value of the affixes. In doing so, they follow the traditional idea that passive markers developed late in the prehistory of Indo-European languages, because of the lack of a common passive marker in these languages. They also assume that this alleged origin of the “passive aorist” could also account for the presence of verb forms with active inflection in passive constructions: the relationship between active inflection and passive constructions would be the outcome of a later process and would not concern the reconstructed Indo-European or unattested stages of Ancient Greek. From this perspective, aorist forms with the affixes -ē-/-thē- should be characterized by active inflection because in the oldest ( unattested) stages of Ancient Greek they did not correlate with passive (about this point, see infra).⁶

The point of view argued here is different. It makes a clean sweep of the semantic interpretations, which have been traditionally attributed to these verb forms, and follows Saussure’s idea that linguistic items have no value by themselves (and therefore do not exist by themselves), but are determined (and created) only through their syntagmatic (or in praesentia) and paradigmatic (in absentia) relationships. Along these lines, I shall try to define the functional values of markers, such as inflection and affixes, that characterize aorist verb forms, particularly “passive aorist” and sigmatic aorist. The outcome will be a picture of the whole aorist system: the functional values of both inflections and affixes could not be accounted for without considering syntactic contrasts such as those illustrated in (2)-(4). The approach suggested here is based on the theoretical and methodological assumptions of La Fauci and Tronci (2009), from which this paper draws its inspiration.⁷

1.5. The analysis of the Ancient Greek data developed here could shed new light on certain diachronic phenomena involving the aorist system, accounting for some morphosyntactic aspects of Modern Greek as well. The following pairs are extracted (with some small changes) from the Modern Greek grammar edited by Holton et al. (1997) and illustrate once again the passive vs. active contrast. Verb forms occurring in these examples belong to the present (5) and the aorist systems (6), respectively:
(5a) oi politikoí agapiountai apó ton kósmo

det-nom.pl politician-nom.pl appreciate-ind.prs.3pl.mid by det-acc.sg people-acc.sg

‘politicians are appreciated by the people’

(5b) o kósmos agapáei tous politikoús

det-nom.sg people-nom.sg appreciate-ind.prs.3sg.act det-acc.pl politician-acc.pl

‘people appreciates politicians’

det-nom.sg first-nom.sg book-nom.sg det-gen.sg Chomsky publish-ind.aor.3sg.act by

det-acc.sg publishing-acc.sg house-acc.sg Mouton

‘the first book by Chomsky was published by the Mouton press’

(6b) o ekdotikós oíkos Moúton demosíeuse to próto biblío tou Tsomski

det-nom.sg publishing-nom.sg house-nom.sg Mouton publish-ind.aor.3sg.act det-acc.sg first-acc.sg book-acc.sg det-gen.sg Chomsky

‘the Mouton press published the first book by Chomsky’
In the present system, inflectional contrast correlates with the syntactic passive vs. active contrast: middle inflection occurs in passive constructions (5a), whereas active inflection characterizes active constructions (5b). Therefore, in this regard there is no difference between Ancient and Modern Greek. However, the aorist system presents a different picture: there is no inflectional opposition, and only active inflection is used in both passive and active constructions. The syntactic difference appears to be expressed by morphemes that linearly precede inflection: the affix -tē(k)- occurs in passive constructions and never occurs in the active ones, in which the sigmatic affix -s- is found, instead.⁸ From the point of view of form, these affixes are related to Ancient Greek -thē- and -sa- (see above).

With respect to the aorist system, an important difference distinguishes Ancient and Modern Greek. In Ancient Greek the value of inflection is not determined: in some morphological classes, it has contrastive value, but in others it does not. Modern Greek has no inflectional contrast, and verb forms exhibit only active inflection. Is there a relationship between the two systems? How could we account for the differences?

2.1. As suggested by La Fauci and Tronci (2009), the passive vs. active contrast appears to be an useful operational tool to test the relationship between forms and functions, in particular between verb forms and their syntactic values. The passive vs. active contrast distinguishes structures that are obviously different at various levels of analysis.⁹ For our purposes here, the relevant level is syntax. Our discussion will feature examples of the active term of constrast drawn only from transitive constructions, in which Subject and Direct Object relations co-occur without any further syntactic process.

In this paper a syntactic structure is conceived as a combination of grammatical relations whose functional value is to be determined according to their syntagmatic and paradigmatic relationships. These relationships are expressed by differences concerning case marking, verb inflection, interpretation and so on. In the framework of Relational Grammar (RG)¹⁰, grammatical relations, such as Predicate, Subject, Direct Object and Indirect Object, are primitives of the syntactic theory. These relations do not pertain to the clause on the whole, but to syntactic strata (the different levels of representation): the final Subject is the Subject of the final stratum (roughly corresponding to the surface representation in GB theory) the initial Object is the Object of the initial stratum (roughly, the equivalent of the underlying representation in GB), and so on. Each of these strata contains all syntagmatic relationships that are relevant for that syntactic stratum; the sequence of strata allows us to identify the paradigmatic relationships that are relevant for each item of the clause. This analysis of syntactic structure, as sequence of strata, allows us to determine the relationships between grammatical relations and morphosyntactic properties, such as verb morphology, the main topic of this paper.

Let us begin with some simple syntactic structures, such as the transitive constructions in all the (b) clauses of (1)-(6). Besides the Predicate, these structures contain two grammatical relations: Subject and Direct Object. These relations are marked in Ancient Greek by the following morphosyntactic properties, whose value is correlative:

1. 
   verb agreement (with the Subject and not with the Direct Object);
   
2. 
   case marking (nominative for the Subject, accusative for the Direct Object¹¹);
   
3. 
   interpretation (active, e.g. ‘those which slew’, ‘those which killed’ for the Subject of (1b), (2b); passive, e.g. ‘those which were slain’, ‘that who was killed’ for the Direct Object of the same constructions).
   

The framework of RG allows us to capture and formalize this difference. Compared with the Subject of transitive structures in (1b)-(2b), the Subject of (1a)-(2a) has a paradigmatic relationship with the Direct Object: this relationship accounts for the interpretations as ‘those which were slain’, ‘that who was killed’, i.e. the same interpretations of the Direct Object of the transitive structures in (1b)-(2b). Passive clauses, which have a more complex syntactic structure than transitive clauses, involve a sequence of strata.¹² The initial syntactic stratum of passive structures is transitive: it is reflected in the interpretation, hypò tôn Perséōn in (1a) having the same interpretation ‘those which slew’ than the Subject of the transitive construction in (1b). The final stratum is an intransitive stratum, because of the advancement of the initial Direct Object to Subject and the “Chômeur Condition”, which says that

veil-dat.sg so over cover-ind.aor.3sg.mid divine-nom.sg goddess-gen.pl

‘and with a veil over all did the fair goddess veil herself’

(8b) pardaléēi mèn prôta metáfrenon eurù kálupse / poikílēi […] (Il. X 29-30)

(9b) [...] hóti Héktora dîon élusa / patrì phílôi [...] (Il. XXIV 593-4)

‘that I have given back noble Hector to his dear father’

then their-dat.sg valour-dat.sg Danaan-nom.pl break-ind.aor.3pl.mid battalion-acc.pl

‘even then the Danaans by their valour brake the battalions’

(10b) Aías dè prôtos Telamṓnios hérkos Akhaiôn / Trṓōn rhêxe phálagga [...] (Il. VI 5-6)

‘Aias, son of Telamon, bulwark of the Achaeans, was first to break a battalion of the Trojans’

Let us start with the pair in (8), focusing on the grammatical relations. The Predicate relation is borne by the same lexical item. Besides the Predicate, in (8a) there is only one other grammatical relation (the Subject), whereas in (8b) we find two argumental relations (Subject and Direct Object). This means that the final stratum of these constructions is intransitive in (8a) and transitive in (8b). The interpretations associated with these grammatical relations indicate that the initial stratum of both constructions is transitive. The initial stratum of (8a) is transitive, as in (8b). In the initial stratum both the Subject and Direct Object relations are borne by the same item in (8a), whereas they are borne by two different items in (8b). The diagrams in (11) show the corresponding representations: a direct reflexive structure in (8a), and the corresponding transitive structure in (8b). The syntactic configuration in (8a), where one nominal bears two grammatical relations in the same stratum (in our example, Subject and Direct Object), is called multiattachment. The notion of multiattachment was introduced by Rosen (1988 [1981]: 43) in her analysis of reflexive clauses in Italian and accounts for several other morphosyntactic phenomena in the Romance languages, as Rosen (1988 [1981]), La Fauci (1984), (1988) and Perlmutter (1989) have discussed. Benedetti (2005) also uses multiattachment in her analysis of Ancient Greek constructions such as those discussed here:

(13) Diagram

To sum up. The various morphological classes of aorist show some differences as far as the inflection contrast is concerned. The sigmatic class is particularly interesting because the inflection contrast does not correlate with the passive vs. active contrast, contrary to what may be observed in the thematic and root classes. However, inflectional contrast is relevant for the sigmatic aorist as well: middle inflection occurs in various kinds of constructions. All of these structures involve a final Subject that bears the Object relation in a syntactic stratum in a 1, 2 multiattachment configuration.

The correlation between middle inflection and multiattachment does not involve only the sigmatic forms. It is also found in the other aorist forms (i.e. thematic and root), and in the present system, as shown by the following pairs:
(14a) amphì d’ ár’ ṓmoisin bálet’ aigída thussanóessan (Il. V 738)

around so then shoulder-dat.pl cast-ind.aor.3sg.mid aegis-acc.sg tasselled-acc.sg

‘around her shoulders she flung the tasselled aegis, fraught with terror’

(14b) amphì dé min phâros kalòn bálon ēdè khitôna (Il. XXIV 588)

around so he-acc.sg cloak-acc.sg fair-acc.sg cast-ind.aor.3pl.act and tunic-acc.sg

‘and had cast about it a fair cloak and a tunic’

grasp-ind.aor.3sg.mid so sceptre-acc.sg paternal-acc.sg imperishable-acc.sg ever

‘and he grasped the sceptre of his fathers, imperishable ever’

(15b) […] sákos heîle tetugménon huîos heoîo (Il. XIV 9)

‘and he grasped the shield of his son’

bury-ptcp.aor.act.nom.pl so det-nom.pl Scythian-nom.pl cleanse-ind.prs.3pl.mid way-dat.sg that-dat.sg

‘after the burial the Scythians cleanse themselves that way’

(16b) autàr épeita thrónous perikalléas ēdè trapézas / […] káthairon (Od. XXII 452-3)

‘When they had done this, they cleaned all the tables and seats’

‘let us not yet loose our single-hoofed horses from their chariots’

(17b) híppous d’Eurumédōn therápōn lúe toîo gérontos / ex okhéōn [...] (Il. XI 620-1)

horse-acc.pl so Eurymedon-nom.sg attendant-nom.sg loose-ind.impf.3sg.act det-gen.sg

old-gen.sg from chariot-gen.pl

‘and Eurymedon the attendant loosed old Nestor’s horses from the chariot’

(18) Table A

3.1. In the aorist system the correlation between passive constructions and middle inflection is restricted to root and thematic verb forms, which are not productive from a morpho-lexical point of view (see examples in (2) above). As a rule, passive clauses exhibit verb forms belonging to the so-called “passive aorist” (see examples in (3a), (4a) above). Morphological markers of this class are the affixes -ē- or -thē-, always combined with active inflection.¹⁸ In spite of this, the syntactic passive vs. active contrast is still manifested by verb morphology. In passive structures, inflection is always combined with the affixes -ē- or -thē-. By contrast, in active structures inflection is never combined with the affixes -ē-/-thē-: it may be combined with an affix, such as the sigmatic affix in (3b) and (4b) above and (19b) below, but this is not necessarily the case, as in (20b), where a root aorist occurs.

det-nom.pl then many-acc.pl slay-ind.aor.afx.3pl.act Hektor-dat.sg noble-dat.sg

‘those many have been slain by goodly Hector’

(19b) khersìn Akhillḗos dámase glaukôpis Athḗnē (Il. XXII 446)

‘flashing-eyed Athene had vanquished him at the hands of Achilles’

‘but the comrades of whom you ask were slain’

(20b) [...] hóte dîon Ereuthalíōna katéktan (Il. IV 318-319)

when noble-acc.sg Ereuthalion-acc.sg slay-ind.aor.1sg.act

‘when I slew goodly Ereuthalion’
As we can see in the (a) examples in pairs (21)-(22), the “passive aorist” does not occur exclusively in passive clauses. In fact, these clauses are intransitive, as shown by the comparison with the corresponding transitive clauses in (b) (see Appendix 4 as well):
(21a) énth’ephánē méga sêma [...] (Il. II 308)

then show-ind.aor.afx.3sg.act great-nom.sg portent-nom.sg

‘then appeared a great portent’

(21b) hēmîn mèn tód’ éphēne téras méga mētíeta Zeùs (Il. II 324)

‘To us has Zeus the counselor showed this great sign’

swift-nom.pl so charioteer-nom.pl gather-ind.aor.afx.3pl.act

‘and the swift charioteers gathered together’

(22b) enthád’ aph’humetérōn políōn ḗgeira hékaston (Il. XVII 222)

meanwhile from your-gen.pl city-gen.pl gather-ind.aor.1sg.act each-acc.sg

‘[...] did I gather each man of you here from your cities’

In both passive and initially unaccusative constructions the final Subject is in a paradigmatic relationship with the Direct Object relation. The difference is that this Object relation belongs to a transitive stratum in passive constructions, but to an unaccusative stratum in (initially) unaccusative constructions. The diagrams in (23) illustrate the syntactic structures of both passive and (initially) unaccusative constructions.

(20b)	hoi dê polees	damen	Hektori diôi		(22b)	mega sêma	ephanê
(21b)	hetaroi	katektathen	unspec.		(23b)	hippêes	agerthen
	1	P	F			1	P
	2	P	1			2	P

With respect to the middle vs. non-middle contrast, “passive aorist” correlates with middle. This relationship accounts for the distribution of “passive aorist”, which differs from that of verb forms having active inflection (without affixes -ē-/-thē-, obviously) and coincides with the distribution of verb forms having middle inflection (see examples in Appendix 5). The distributional differences are obvious when we compare constructions in (c) below with the (a) examples in § 3.1 above, repeated below for convenience.²⁰

det-nom.pl then many-acc.pl slay-ind.aor.afx.3pl.act Hektor-dat.sg noble-dat.sg

‘[...] those many have been slain by goodly Hector’

(19c) hṑs ára puknà karḗath’huph’Héktori dàmnato laôn (Il. XI 309)

‘just so many heads of men were laid low by Hector’

‘but the comrades of whom you ask were slain’

(20c) [...] allá poth’ hôde kataktenéesthe kaì húmmes (Il. XIV 481)
but then that way slay-ind.fut.2pl.mid and you-nom

‘but some day in like manner you too will be slain’

then show-ind.aor.afx.3sg.act great-nom.sg portent-nom.sg

‘then appeared a great portent: a serpent, blood-red on its back, terrible’

(21c) [...] hékathen dé te phaínetai augḗ (Il. II 456)

‘and from afar can the glare be seen’

swift-nom.pl so charioteer-nom.pl gather-ind.aor.afx.3pl.act

‘and the swift charioteers gathered together’

(22c) allà prìn epì éthne’ageíreto muría nekrôn (Od. XI 632)

‘but so many thousands of ghosts gathered and uttered such appalling cries’

3.3. Let us consider now the relationship between forms and functions in the most productive classes of the Ancient Greek aorist system, that is, “passive aorist” and sigmatic aorist. Both the “passive aorist” and the sigmatic aorist correlate with the middle term of the middle vs. non-middle constrast. Nevertheless, this relationship is expressed by means of different markers: the affixes in the “passive aorist”, and inflectional endings in the sigmatic aorist.

Although it is true that both verb forms correlate with the middle (vs. non-middle), it is also true that they do not occur in the same constructions, as argued by Benedetti (2005) and Tronci (2005). Both verb forms correlate with a Subject relation that is in a paradigmatic relationship with the Direct Object relation. Nevertheless, this paradigmatic relationship involves different syntactic processes: the Subject relation occurring with the sigmatic aorist combines with the Direct Object relation in a multiattachment-stratum, whereas no multiattachment is involved in constructions having the “passive aorist”.

Nevertheless, the “passive aorist” and the sigmatic aorist appear to occur as free variants in a small group of clauses, as illustrated in examples (24)-(25).²¹
(24a) kaì tóte dḕ perì kêri Poseidáōn ekholṓthē (Il. XIII 206)
and then so at heart-dat.sg Poseidon-nom.sg move to rage-ind.aor.afx.3sg.act

‘And then it was that Poseidon grew furious at heart’

(24b) [...] potamòs dè kholṓsato kēróthi mâllon (Il. XXI 136)

river-nom.sg so move to rage-ind.aor.3sg.mid heart-loc.sg still

‘and the river grew still more angry within’

‘we camped upon the beach’

(25b) hoi mèn koimḗsanto katà mégara skióenta (Od. X 479)

they-nom so get to sleep-ind.aor.3pl.mid in cloister-acc.pl covered-acc.pl

‘the men laid themselves down to sleep in the covered cloisters’

‘just as but now the lord of men, Agamemnon, moved me to rage’

(25c) tòn d’ autoû koímēse Gerḗnios hippóta Néstōr / Tēlémakhon, [...] (Od. III 397-8)

he-acc so there get to sleep-ind.aor.3sg.act Nestor-nom.sg Telemachus-acc.sg

‘but the horseman, Nestor of Gerenia, put Telemachus to sleep [in the room]’

My hypothesis is that the “passive aorist” and the middle sigmatic aorist signal a syntactic difference even in constructions which appear to be similar to each other, such as those in (24a-b) and (25a-b). This difference may be accounted for in terms of the presence vs. absence of multiattachment involving the domain of middle structures. Both (a) and (b) in (24)-(25) contain an initially unaccusative stratum: for this reason, they show the same interpretation. The difference involves the syntactic processes affecting the initial Direct Object relation and the final Subject relation: whereas in (24b) and (25b) there is multiattachment (with the same nominal bearing both the Subject and Object relations), in (24a) and (25a) there is no multiattachment. The diagrams in (26) illustrate the contrast in structure:

My analysis suggests that syntactic processes in Ancient Greek are independent from semantics. In fact, the analysis shows that morphosyntactic phenomena such as those involving verb mophology are associated with syntactic processes that sometimes do not have any semantic correlates.²³ The analysis is also interesting from a crosslinguistic perspective because it captures important functional similarities between Ancient Greek and Italian morphosyntax in spite of the obvious differences in form.

4.1. Let us now move from Ancient to Modern Greek. As already noted above, the expression of the passive vs. active contrast correlates with inflectional contrast in the present system, whereas in the aorist system it is marked by the affixes -tê(k)- and -sa-, respectively. These affixes developed from the Ancient Greek affixes -thē- and -sa-. The affix -tē(k)- is found in verb forms that occur not only in passive constructions, but also in other kinds of middle constructions, such as initially unaccusative and reflexive structures, as shown by clauses in (27a) and (28a). The examples in (27b) and (28b) show the corresponding transitive constructions.²⁴

now when det-nom.pl Ithacan-nom.pl assemble-ind.aor.afx.3pl.act and all together

meet-ind.aor.afx.3pl.act

‘now when they were assembled and met together’

(27b) ki ekeínoi élegan ti tous ḗthelan kai tous monosunáxan (Od. III 140)

and those-nom.pl say-ind.impf.3pl.act why they-acc want-ind.aor.3pl.act and they-acc assemble-ind.aor.3pl.act

‘and they said their say, and told why they had gathered the host together’

with two hands after with oil-acc.sg anoint-ind.aor.afx.3sg.act soft-acc.sg ambrosial-acc.sg

‘she anointed herself richly with oil, ambrosial, soft’

(28b) […] kai t’áleipse me ládi ploúsia […] (Od. XIX 505)

‘and she anointed him richly with oil’

anoint-ind.aor.3pl.mid so richly oil-dat.sg ambrosial-dat.sg soft-dat.sg

‘she anointed herself richly with oil, ambrosial, soft’

(29b) […] kaì ḗleipsen líp’elaíōi (Od. XIX 505)

‘and she anointed him richly with oil’

This functional relationship with middle is marked by the affixes -ē-/-thē-, which (as we have already seen) are always combined with active inflection. Obviously, the value of the active inflection found in these verb forms is different from the active inflection which is found in other verb forms and which contrasts with middle inflection. Since the opposition between middle and active inflection correlates with a syntactic difference, active inflection has contrastive value. On the other hand, active inflection that occurs in combination with the affixes -ē-/-thē- has not contrastive value, because it does not contrast with middle inflection. In this case, inflection registers agreement in person and number with the Subject, and does not express other properties of the Subject. Therefore, in the “passive aorist” the relationship between function and form, as far as the affix and the inflection are concerned, is reversed in comparison to the sigmatic aorist, which otherwise seems to be morphologically similar to the “passive aorist”. In both classes, indeed, inflection combines with an affix. The difference is that the affixes -ē-/-thē- have contrastive value: the verb forms that exhibit them occur in syntactic structures that are middle [without multiattachment] and do not occur in both active and middle [multiattachment] constructions. By contrast, the affix -sa- does not have any contrastive value: verb forms that exhibit this affix may occur in both active and middle [multiattachment] constructions, this functional difference being manifested by inflectional contrast .

Therefore, in spite of the morphological similarity, the sigmatic and “passive aorist” classes are very different as far as the relationship between forms and functions is concerned. Inflection that combines with affix -sa- has contrastive value: it correlates with the opposition middle [multiattachment] vs. non-middle. Inflection that combines with affix -thē- has not contrastive value. Table C summarizes the relationship between functional values and inflection:
(31) Table C
As Table C shows, active inflection is the unmarked term of the opposition, because it occurs in both non-middle and middle constructions. Middle inflection, on the other hand, is the marked term of the opposition: it occurs only in middle constructions.

The unmarkedness of active inflection correlates with the non-contrastive value of inflection and, concurrently, with the contrastive value of affixes, as is represented in the first row of Table C. The affixes -sa- and -thē- are markers of the non-middle vs. middle opposition: they have contrastive value.

Seen from this perspective, the loss of verb forms that exhibit middle inflection appears to be a consequence of their markedness. By comparison with other forms in Table C, the middle sigmatic aorist is marked with respect to the relationship between functions and forms. Two morphemes occur in it: the affix -sa- (with non-contrastive value) and the middle inflection (with contrastive value). The unmarkedness of active inflection correlates with its non-contrastive value, whereas the unmarkedness of affixes correlates with their contrastive value. With respect to both these correlations, the middle sigmatic aorist is marked.

These different relationships between forms and functions appear to account for the diachronic changes affecting these verb forms. The loss of the middle inflection in Modern Greek represents the outcome of the clash between the two different systems coexisting in Ancient Greek, that is, a system in which the inflection opposition has contrastive value, and another one in which the affix opposition has contrastive value.

I do not rule out the possibility that morphological factors may be involved in the process. Such factors may include, for instance, the formal similarity of the sigmatic aorist and the “passive aorist”, both of which combine an affix with inflection:

(32a) […] allà phóbēthen / pántes […] (Il. XVI 659-660)
but drive in rout-ind.aor.afx.3pl.act all-nom.pl

‘but [they] were driven in rout one and all’

(32b) […] bóes hṓs, / hás te léōn ephóbēse […] (Il. XI 172-3)

cow-nom.pl like rel-acc.pl and lion-nom.sg drive in rout-ind.aor.3sg.act

‘like cattle that a lion has driven in rout’

(33a) autàr epeì kósmēthen hám’hēgemónessin hékastoi (Il. III 1)

but when prepare-ind.aor.afx.3pl.act with leader-dat.pl each-nom.pl

‘now when they drew up, the several companies with their leaders […]’

(33b) allá spheas kósmēse Podárkēs ózos Árēos (Il. II 704)

but they-acc prepare-ind.aor.3sg.act Podarces-nom.sg scion-nom.sg Ares-gen sg

‘but Podarces, scion of Ares, marshaled them’

(34a) […]ê méga érgon huperphiálōs etelésthē /Tēlemákhōi hodòs hēde […] (Od. IV 663-4)

indeed great-nom.sg work-nom.sg superbly fulfill-ind.aor.afx.3sg.act Telemachus-dat.sg voyage-nom.sg this-nom.sg

‘this voyage of Telemakhos is a very serious matter’

(34b) […] sù dé hoi nóon ouk etélessas (Il. XXIII 149)

you-nom he-dat intent-acc.sg not fulfill-ind.aor.2sg.act

‘but you did not fulfill for him his intent’

The reasons for the diachronic change are to be found in markedness relationships involving the forms and functions of markers, such as inflection and affixes, and their combinations. Consequently, the diachronic process is to be conceived as a neutralization of the contrastive value of inflection whose end result is the loss of forms featured by this contrastive value (i.e. the middle sigmatic forms).

In this paper I have provided a descriptive characterization of the relationship between forms and functions in the aorist and the present systems of Ancient Greek. This description clearly shows that some aspects of the synchronic relationships relevant to the aorist system of Ancient Greek correlate with some diachronic processes leading to the Modern Greek aorist system. The comparison between the two aorist systems points out an important difference: whereas in Ancient Greek there are aorist verb forms that exhibit middle inflection, no aorist verb form characterized by middle inflection is found in Modern Greek. In spite of this difference, the two systems show an evident continuity with respect to form: the most productive aorist classes of Ancient Greek include the affixes -thē- and -sa-, which are in an etymological relationship with the affixes -tê(k)- and -sa- in Modern Greek.

In order to account for the difference in inflectional contrast in the aorist system – a contrast that is relevant to Ancient Greek but not to Modern Greek – I have studied the combinations of various verb markers (inflection and affixes) and the different syntactic structures in which the relevant verb forms occur. More specifically, my analysis has explored the different values of these verb forms (and specifically, of the different markers exhibited by them) in terms of their syntagmatic and paradigmatic relationships.

The picture of the Ancient Greek aorist system that emerges from the analysis may account for the diachronic changes affecting this system, changes that may be explained in terms of markedness synchronic relationships between forms and functions.²⁷