Morphological interpretations of syncretism in the panorama of Greek

3.3.1 Masculine nouns ending in -as, -is

In contrast to CG, in MG, masculine nouns ending in nominative -as or -is have an obligatory syncretic inflectional paradigm.

In the singular sub-paradigm, a single form realizes the cells of the genitive, accusative, and vocative cases (see taˈmias, piiˈtis, and anðriˈandas in Table 5). Similarly, in the plural sub-paradigm, a single form realizes the cells of the nominative, accusative, and vocative (see ταμίας, ποιητής, and ἀνδριάς in Table 5). The extensive syncretism that occurs in the nouns of this category results synchronically in only the nominative cell for the singular and the genitive cell for the plural being distinguished morphologically.

3.3.2 Masculine nouns ending in -os

Masculine nouns ending in -os constitute the only group of nouns that, compared to the others, has strongly resisted syncretism over time. In MG, these masculine nouns display the greatest morphological differentiation among the cells of the paradigm. Except for the loss of the dative case and changes in the phonological system of Greek, the group of masculine nouns in -os retains an inflectional paradigm that has remained unchanged through time.

The diachronic stability of the inflectional class of masculine nouns in -os is the only exception among all MG inflectional classes. This single peculiarity should probably be attributed to the reduced productivity of the class of nouns in -os for MG. Indeed, as Marinis (2015) has shown, the class of masculine nouns in -os rarely incorporate new members and integrate few loanwords. Thus, the reduced productivity of this inflectional class has contributed to its greater resistance to language change compared with other classes. This conservatism partly accounts for the inflectional class of masculine nouns in -os failing to accord with the rest of MG nouns in developing extensive syncretism.

It should be noted that the resistance of the class of masculine nouns ending in -os to syncretism is not confirmed in all MGD’s. In some of them,^[5] which I call MGD(2), the phenomenon is attested in the plural of masculine nouns in -os.

Italiot and the Tsakonian^[6] of the Peloponnese may be considered to present another case of nominative-accusative syncretism, restricted exclusively to the singular sub-paradigm. However, the behavior of the forms in these dialects should not be considered to represent an endogenous tendency of their paradigmatic inflection but rather to be a byproduct of an extra-paradigmatic, general phonetic change. In particular, as Marinis (2020) has shown, the silencing of the ending segments -s and -n in these dialects has stimulated a morphological change leading to a syncretism of the nominative and accusative cells in the singular sub-paradigm of these nouns (see, for example, the paradigm ˈpono, in Table 6).

4.2.1 The sub-paradigm of the singular

Syncretism between the accusative and the genitive is minimally present, if at all, in MG nominal morphology,^[7] while it is completely absent in CG. In SMG, the phenomenon is restricted exclusively to the class of masculine nouns ending in -as and -is (see Table 5). Even in this inflectional class, however, the phenomenon occurs only in the singular sub-paradigm and never in that of the plural.

The syncretism of genitive and accusative in the class of masculine nouns in -as and -is must be interpreted as the product of an extensive intra-paradigmatic uniformity. Since the segment -a- already existed in antiquity in all the cells of the singular sub-paradigm except the cell realizing the genitive case, it also passed into the cell of the genitive singular through paradigmatic leveling. The process was triggered by the need for the system to have a 1:1 correspondence between meaning and function. Here, this meant reducing the amount of stem allomorphism available in the singular sub-paradigm. The described paradigmatic leveling was facilitated by the reanalysis that -a- underwent at some point in its history so that it is now considered part of the stem (see Marinis to press; Marinis 2020; Ralli 2022). Thus, the cell of the genitive singular ceased to be morphologically realized with the suffix -u, whose use was specialized for names in -os. Today, in the class of masculine nouns in -as and -is, the genitive singular is realized with -a or -i, respectively, due to intra-paradigmatic leveling.

4.2.2 The sub-paradigm of the plural

Upon careful examination of all the data from MG, as shown in Tables 3 –6, one observes that, in MG, the syncretism between the accusative and the genitive cells is utterly absent from the sub-paradigm of the plural. This is true for all the inflectional classes available in the system, and in the data shown in Tables 3 –6, it is without exception. At the same time, in the singular sub-paradigm, accusative-genitive syncretism is possible but restricted exclusively to the inflectional classes of masculine nouns ending in -as and -is.

However, for some MGD’s, a very interesting phenomenon has been observed regarding the genitive plural cell of masculine nouns. The phenomenon has been extensively discussed for Cypriot (Cy.), but it also occurs in other linguistic varieties of Greek.^[8] For the purposes of this paper, let us focus our interest on Cy. as an exemplar for which we have good long-term documentation. Note examples 1–4 below, in each of which a structure is first given in Cy., as it occurs in sources, and then is contrasted with SMG:

In all the above utterances (1-4) from Cy., there is an accusative where we would expect a genitive in the corresponding utterance in SMG. Indeed, as illustrated in (1), the phenomenon is not recent; it has been present in Cypriot since at least the late 19th century.

Based on utterances 1-4, the main question is the morphological nature of the structures marked with bold typeface. Earlier literature (notably Mertyris 2012; Symeonidis 2006) treats forms such as those in 1–4 as genitives. They imply that the cell of the plural genitive is syncretic with that of the plural accusative. For this to be true, at least the four criteria I propose in 5 would have to be met simultaneously:

Thus, interpreting the phenomenon as a syncretism between the genitive and accusative cells of the plural would be particularly problematic, primarily because this analysis would violate the criteria 5b, 5c, and 5d. Specifically, in all the utterances 1–4 cited above, the article used is that of the accusative plural (ˈtus), even though the article of genitive plural (ˈton) is available in the dialect. Moreover, the earlier analysis does not take into account the fact that in the same sources from which examples 1–4 are drawn, a parallel ‘second form’ frequently exists, representing the plural genitive. This second form is morphologically marked as a plural genitive (6, and CyP_b in Table 8).

Furthermore, a close examination of 1–4 as a genitive-accusative plural syncretism would raise serious questions about the methodological consistency and theoretical elegance of the analysis. For example:

1. What causes the systematic syncretism (assuming that it is, indeed, syncretism) not only of the noun but also of the article?

2. Why does this phenomenon occur only in masculine nouns but not feminine and neuter nouns for which the plural cell is absent?

3. What is the factor that causes the alleged syncretism among the cells of genitive and accusative in masculine nouns?

I maintain that what really causes phenomena like those appearing in examples 1–4 is the general weakness of the cell for the genitive plural in MG. The genitive plural cell seems to be in retreat,^[9] as evidenced, among other things, (a) by the inability of native speakers to determine the position of the stress (i.e., 'anθropu and/or anθ'ropu ‘human._GEN.SN.’; see Sims’ [2015] extensive analysis), and (b) by the defectiveness of the diminutive -aci (and some other lexical categories) for the genitive plural.^[10] For example, although based on its inflectional class (IC), the word mihaˈnaci ‘motorbike’ should form the genitive singular *mihanaˈciu and the genitive plural *mihanaˈcion,^[11] these cells in fact cannot be realized, so the inflectional paradigm remains defective.

A more tenable hypothesis to explain these phenomena would be that the failure of the genitive plural to be realized allowed the less marked case (i.e. the accusative) of the plural sub-paradigm to express the functions of the genitive by default. Genitive still exists morphologically in the Cypriot system but is problematic for speakers.

Moreover, if indeed, in examples 1–4, there were a syncretism between the genitive and the accusative plural, we would expect it to be in the form CyP_c (Cypriot Paradigm, Table 8) and not in the form CyP_a (Table 8), as reported by Symeonidis (2006), Mertyris (2012), etc.

If the analysis I propose above is correct, the following question arises:

How is it possible that the cell of the genitive plural, the structural weakness of which has been noted repeatedly,^[12] is not syncretized and, indeed, remains the only cell of the nominal paradigm of the Greek language that resists any syncretism?

The phenomenon does not appear to be coincidental. To understand what is happening, we need to consider a paradigmatic peculiarity of the genitive plural cell not usually mentioned in the literature. Consider Table 9, which lists all the inflectional suffixes of the nominal system of SMG, by case and by IC, based on the classification of nouns proposed by Ralli (2022: 163–165).

A comparative examination of the inflectional suffixes of SMG makes it clear that the cell of the genitive plural is the only one in the entire nominal inflectional system marked inter-paradigmatically with the same inflectional suffix, namely -on. I believe this is the reason the genitive plural cell is never syncretized in Greek. However, at times, the genitive plural cell is not realized at all (cf.-aci), but it still never participates in syncretism with other paradigmatic cells.^[13]

5.3.1 The role of IC

This dependence of syncretic patterns on the lexeme’s IC is relevant only for morphology and not for any other level of linguistic analysis. The dependence of syncretism on the particular IC should, therefore, be considered a morphomic ^[21] phenomenon. As is shown in subsection 3.3.2, the only IC of SMG that comes close to completely resisting syncretism is that of masculine nouns in -os, which has very little productivity as it is marked as [+learned], and is applied to only a few loanwords.

5.3.2 The role of inter-paradigmatic similarity

At the same time, the resistance of the cell representing the genitive plural to engage in any syncretism cannot be entirely explained by the underlying semantic features of cases (see Section 5.2); rather, it can better be accounted for by paradigmatic morphology. As shown for SMG in subsection 4.2.2, the failure of the cell of the genitive plural to be syncretized into any of the inflectional classes available in the system occurs because the cell for the genitive plural is always inter-paradigmatically marked with the same inflectional suffix. It is no coincidence that, of all the cells in the SMG inflectional paradigm, the genitive plural is the only one that:

1. is never syncretized, and

2. carries the same suffix inter-paradigmatically (see Table 9).

Based on these observations from the nominal inflectional system of Greek, I propose a decisive new inhibitory factor to account for the emergence of syncretism:

1. The inter-paradigmatic similarity of a cell, achieved through the inter-paradigmatic realization of the cell with the same inflectional suffix, prevents this particular cell from participating in (intra-paradigmatic) syncretism.

This mechanism works exceptionally well for Greek, but the extent of a broader application needs to be tested by further typological research.

This story becomes more intriguing if one considers that the cell of the genitive plural is the weakest in the entire nominal system of Greek.^[22] Thus, we would expect for the genitive plural to be the cell most easily participating in syncretic patterns. Notably, instead of this, the genitive plural cell is easier to be defective^[23] (see, for example, Table 10) than participating in syncretism since the latter phenomenon is prohibited by the inter-paradigmatic realization of the cell with the same inflectional suffix, namely -on.

5.3.3 A highly unnatural case of syncretism

Finally, as noted by Marinis (2020), although historically triggered by phonology, the most unnatural^[24] syncretism among the cells of the genitive singular and accusative plural in what Marinis defines as IC1 of Italiot (cf. ponu ‘pain’._{GEN.SN./ACC.PL.}), still synchronically creates a particular kind of syncretism which also has a morphological character.

From standpoint, I argue that even if a phenomenon diachronically has its roots in phonology, at a synchronic level, it is still entirely possible to create associations in the system that, for the native speaker, no longer have phonological explanations. The accumulation of such associations, which are no longer phonologically interpretable, can, in turn, cause the restructuring of the inflectional system through the mechanisms of intra-paradigmatic and inter-paradigmatic uniformity.^[25]