Next Article in Journal
The Influence of Language Experience on Speech Perception: Heritage Spanish Speaker Perception of Contrastive and Allophonic Consonants
Previous Article in Journal
Pragmatic Perception of Insult-Related Vocabulary in Spanish as L1 and L2: A Sociolinguistic Approach
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Breaking Down Greek Nominal Stems: Theme and Nominalizer Exponents

by
Giorgos Markopoulos
Department of Mediterranean Studies: Archaeology, Linguistics, International Relations, University of the Aegean, 85132 Rhodes, Greece
Languages 2025, 10(4), 85; https://doi.org/10.3390/languages10040085
Submission received: 28 January 2025 / Revised: 7 April 2025 / Accepted: 9 April 2025 / Published: 17 April 2025

Abstract

:
This article focuses on the right edge of nominal stems in Greek and aims to show that stem-final segments should be analyzed as distinct morphological constituents. Two types of such constituents are identified. On the one hand, stem endings such as -a(ð), -i(ð), and -a(t) have a predictable distribution, as they are found in nouns with specific morphosyntactic properties and stress patterns. On the other hand, stem endings like -o, -a, and -i cannot function as predictors of the morphosyntactic status of the noun, although they may convey information about its stress position. The distinction between the two constituent categories is captured through an analysis couched within Distributed Morphology. Specifically, it is proposed that stem endings of the first category function as nominalizer exponents, while those of the second category serve as exponents of a Theme node, which is inserted post-syntactically and bears no grammatical features. The allomorphic variation exhibited by these exponents is accounted for by means of a phonological analysis based on Gradient Harmonic Grammar. The proposed approach is shown to capture empirical generalizations that have been overlooked in traditional grammatical descriptions and theoretical analyses based on multiple stem allomorphs.

1. Introduction

Greek nominal morphology is mostly fusional. Nominal forms usually comprise two main morphological constituents: (a) a stem, which, apart from bearing the semantic content of the noun, is also considered to be specified for grammatical gender (masculine/feminine/neuter; see, among others, Ralli, 2002, 2022; Anastassiadis-Symeonidis & Cheila-Markopoulou, 2003); (b) an inflectional suffix, which expresses number (singular/plural) and case (nominative/accusative/genitive/vocative) distinctions (e.g., Triantafyllidis, 1941/2018; Clairis & Babiniotis, 2004; Holton et al., 2012; Ralli, 2022; Markopoulos, 2023). An indicative example is given in (1):
(1)a.ánθropo-s‘human.MASC-SG.NOM’
In this article, we focus on the right edge of Greek nominal stems, which poses several theoretical challenges.1 First, it is the only part of the nominal form that exhibits allomorphy, with the stem-final vowel in the singular form either being absent (2) or followed by the consonant /ð/ or /t/ (3) in the plural form:2
(2)a.pórta-Ø‘door.FEM-SG.NOM/ACC’3
b.pórt-es‘door.FEM-PL.NOM/ACC’
(3)a.psará-s‘fisherman.MASC-SG.NOM’
b.psaráð-es‘fisherman.MASC-PL.NOM/ACC’
The second theoretical challenge comes from the fact that the allomorphy pattern of a nominal stem is often not predictable: multiple stems may share the same ending, but, nevertheless, they may be associated with different properties or behavior with respect to allomorphy. This is evident in the examples in (4–6), where, although all three nouns share the same stem-final segment /a/ in the SG.NOM form, they vary in several aspects. Specifically, each one has a different grammatical gender value ((4): FEM; (5): MASC; (6): NEUT), allomorphy pattern ((4): -a/-Ø; (5): -á/-áð; (6): -a/-at), and stress pattern ((4): paroxytone; (5): oxytone; (6): proparoxytone). Thus, it seems that the morphosyntactic and phonological properties of a nominal stem ending in /a/ cannot be straightforwardly predicted.
(4)a.karékla-Ø‘chair.FEM-SG.NOM/ACC’
b.karékl-es‘chair.FEM-PL.NOM/ACC’
(5)a.babá-s‘dad.MASC-SG.NOM’
b.babáð-es‘dad.MASC-PL.NOM/ACC’
(6)a.máθima-Ø‘lesson.NEUT-SG.NOM/ACC’
b.maθímat-a‘lesson.NEUT-PL.NOM/ACC’
On the other hand, one could not ignore that stem-final segments often do function as reliable predictors of certain features of the noun. For instance, all nouns exhibiting the allomorphy pattern -a/-at (6) are neuter and are stressed on the antepenultimate syllable in the SG.NOM/ACC form (provided that they are minimally trisyllabic). Conversely, all nouns exhibiting the allomorphy pattern -á/-áð (5) are non-neuter (i.e., FEM/MASC) and are stressed on the stem-final vowel /a/. This suggests that the right edge of the nominal stem is a crucial morphological unit that may determine the morphosyntactic and phonological properties of the noun.
In this article, we argue that this is indeed the case, namely, that stem endings are individual morphological items with distinct grammatical features. More precisely, adopting the theoretical framework of Distributed Morphology (DM; Halle & Marantz, 1993), we claim that stem endings are exponents of two different types of syntactic constituents: Theme nodes and nominalizers. Under this view, the conundrum posed by examples like the ones presented in (4–6) is actually trivial: although the three nouns appear to have the same stem-final segment at the surface, in reality, each one includes a different Theme/nominalizer exponent.
The remainder of the article is organized as follows: In Section 2, we focus on six stem endings, namely -o, -a, -i, -a(ð), -i(ð), and -a(t), and we examine their distribution in Greek nouns. In Section 3, after a brief overview of the DM framework, we put forth an analysis that highlights the morphosyntactic differences among the six selected stem endings and accounts for their distribution. Section 4 aims to explain why these stem endings exhibit allomorphic variation by presenting a phonological analysis within the framework of Gradient Harmonic Grammar (GHG; Smolensky & Goldrick, 2016). In Section 5, we discuss the advantages of the proposed analysis compared with alternative approaches that account for the allomorphic phenomena in question by positing multiple stem variants for each noun. Section 6 wraps up the article with some concluding remarks.

2. Distribution of Stem Endings

2.1. Stem-Final /o/, /a/, /i/

Let us begin with nominal stems ending in -o by examining the examples in (7):
(7)a.ðáskalo-s‘teacher.MASC-SG.NOM’cf. PL: ðáskal-i
b.uranó-s‘sky.MASC-SG.NOM’cf. PL: uran-í
c.ípiro-s‘continent.FEM-SG.NOM’cf. PL: ípir-i
d.leofóro-s‘avenue.FEM-SG.NOM’cf. PL: leofór-i
e.átomo-Ø‘person.NEUT-SG.NOM/ACC’cf. PL: átom-a
f.biskóto-Ø‘biscuit.NEUT-SG.NOM/ACC’cf. PL: biskót-a
We notice that nouns with stem-final /o/ may bear any grammatical value (masculine (7a–b), feminine (7c–d), or neuter (7e–f)) and may have both human (7a,e) and non-human referents (7b,c,d,f). Furthermore, they may be stressed on any permissible syllable position (ultimate/U (7b), penultimate/PU (7d,f), or antepenultimate/APU (7a,c,e)).4 In other words, there are no morphosyntactic, semantic, or phonological criteria that could conclusively account for the distribution of stem-final /o/ in Greek nouns.
A similar picture emerges when we look at nominal stems ending in -a and -i and exhibiting the allomorphy pattern -a/-Ø and -i/-Ø, respectively. Some representative examples are provided in (8–9).
(8)a.ðaskála-Ø‘teacher.FEM-SG.NOM/ACC’cf. PL: ðaskál-es
b.aɣorá-Ø‘market.FEM-SG.NOM/ACC’cf. PL: aɣor-és
c.fílaka-s‘guard.MASC-SG.NOM’cf. PL: fílak-es
d.kanóna-s‘rule.MASC-SG.NOM’cf. PL: kanón-es
(9)a.aðerfí-Ø‘sister.FEM-SG.NOM/ACC’cf. PL: aðerf-és
b.záxari-Ø‘sugar.FEM-SG.NOM/ACC’cf. PL: záxar-es
c.aθlití-s‘athlete.MASC-SG.NOM’cf. PL: aθlit-és
d.planíti-s‘planet.MASC-SG.NOM’cf. PL: planít-es
It is evident that, in this case too, the morphosyntactic, semantic, and phonological properties of the nouns in question cannot be predicted according to the stem-final segment. Contrast, for instance, nouns that refer to humans in (8a,c) and (9a,c) with nouns referring to inanimate/abstract entities in (8b,d) and (9b,d). Similarly, with respect to stress position, we encounter oxytone (8b; 9a,c), paroxytone (8a,d; 9d), and proparoxytone nouns (8c; 9b). The only difference is that, unlike the examples in (7), the stem endings exemplified in (8–9) can belong to only feminine or masculine nouns.5
Apart from their unpredictable distribution, another common aspect among the stem-final segments /o/, /a/, and /i/ is that they share an overarching allomorphy pattern: the stem-final vowel that appears in the singular form is absent from the plural form. This pattern is consistent independently of the quality of the vowel, the grammatical gender value, and the stress position. This final point—i.e., the fact that the allomorphy pattern remains intact regardless of whether the stem-final vowel bears stress—is particularly important, since, on purely phonological grounds, we would expect that a stressed vowel would be more resistant to deletion compared with an unstressed one.
To conclude, the stem-final vowels /o/, /a/, and /i/ may be grouped together on the basis of their similar behavior with respect to both their distribution and the allomorphy pattern they follow, as summarized in Table 1.

2.2. Stem-Final /a(ð)/, /i(ð)/, /a(t)/

There are many masculine nouns that, despite having the same stem-final segments as the nouns presented in (8c,d) and (9c,d) in the singular, exhibit a different allomorphy pattern, as illustrated in the following examples.
(10)a.vasiljá-s‘king.MASC-SG.NOM’cf. PL: vasiljáð-es
b.papá-s‘priest.MASC-SG.NOM’cf. PL: papáð-es
(11)a.manávi-s‘grocer.MASC-SG.NOM’cf. PL: manávið-es
b.serífi-s‘sheriff.MASC-SG.NOM’cf. PL: serífið-es
Contrary to what we saw in Section 2.1, the nouns in (10–11) retain their stem-final vowel throughout their inflectional paradigm. In fact, in plural forms, the vowel /a/ or /i/ is no longer aligned with the right edge of the stem, since it is followed by the consonant /ð/.
This is not the sole difference between the nouns in (8–9) and those in (10–11). While the former category comprises human and non-human nouns of masculine and feminine gender, the distribution in the latter category is significantly more restricted. In particular, the allomorphy patterns á/-áð and -i/- are predominantly found in masculine nouns with human referents.6 Furthermore, the position of stress in these nouns is totally predictable. In nouns like vasiljás and papás, the stem-final vowel /a/ is always stressed, and, in nouns with stem-final /i(ð)/, stress always falls on the syllable that precedes /i(ð)/ (i.e., on PU in the singular and on APU in the plural).7
Moving on to another category of stems ending in -a, let us examine the examples in (12).
(12)a.pírama-Ø‘experiment.NEUT-SG.NOM/ACC’cf. PL: pirámat-a
b.máθima-Ø‘lesson.NEUT-SG.NOM/ACC’cf. PL: maθímat-a
We observe that these nominal stems behave similarly to the ones given in (10–11), in the sense that they all retain the stem-final /a/ both in the singular and plural forms. What is different here is the consonant that appears at the right edge of the stem in the plural (/t/ instead of /ð/). Moreover, the grammatical gender of the noun is always neuter, and the referent is non-human. Finally, stress falls consistently on the APU syllable.8
Table 2 summarizes the main properties of stem-final /a(ð)/, /i(ð)/, and /a(t)/.
The three stem-final segments/sequences in Table 2 stand in sharp contrast with those in Table 1. On the one hand, /o/, /a/, and /i/ are characterized by random distribution and a consistent V(owel)-Ø allomorphy pattern. On the other hand, /a(ð)/, /i(ð)/, and /a(t)/ have a systematic distribution and an allomorphy pattern that can be formalized as V-VC.
Table 3 illustrates the main points of difference between the two categories of stem endings.

3. Morphosyntactic Analysis

3.1. Distributed Morphology

A theoretical model of morphological analysis that enables us to capture the differences between the two categories of stem-final segments/sequences presented above is DM (Halle & Marantz, 1993, 1994; Harley & Noyer, 1999; Embick & Noyer, 2007; Siddiqi, 2010, 2018, among others). DM does not endorse the lexicalist view that words and phrases are built by separate grammatical components; instead, it postulates a “syntax-all-the-way-down” word derivation. As a realizational framework, it dissociates the grammatical information of a word from its phonological content. This means that, according to DM, the derivation of a word involves two distinct stages: first, syntax creates an abstract hierarchical structure; then, this structure is realized by phonological material and undergoes phonological processing. Crucially, due to language-specific PF requirements, further modifications to the syntactic output may take place between the stages. We will come back to this issue below and in Section 3.3.
A fundamental theoretical assumption of DM is that there are no stems stored in the lexicon. What is traditionally considered to be a stem results from the merge of an acategorial root (a root with no grammatical features whatsoever) and a functional head that assigns to the root a specific grammatical category. For instance, a “nominal stem” is formed through the merge of a root and a nominalizer (n), as illustrated in (13):9
(13)Languages 10 00085 i001
The nominalized root acquires additional grammatical features through the merge with non-categorizing functional heads such as Number (Num).
(14)Languages 10 00085 i002
For Greek, we assume that the Num node, apart from features than encode number distinctions, may be also specified for case features (Alexiadou et al., 2007; Markopoulos, 2018), which are inserted post-syntactically (Embick & Noyer, 2007). In line with previous DM analyses (Kihm, 2005; Lowenstamm, 2008; Kramer, 2015; for Greek, see Alexiadou, 2017; Markopoulos, 2018; Panagiotidis, 2018; Alexiadou & Anagnostopoulou, 2023), we further posit that grammatical gender is hosted on the n node. More specifically, drawing on Kramer’s (2015) and Markopoulos’s (2018) analysis (see also Alexiadou & Anagnostopoulou, 2023), we view grammatical gender distinctions as different ways of nominalizing a root or, in Kramer’s (2015) terms, as different n “flavors”. Some n flavors include the feature [±human] in their featural specification and, therefore, are semantically interpretable, whereas others assign gender values based on strictly formal criteria. A list of representative examples is given in (15):10
(15)a.n[+human, FEM]
b.n[FEM]
c.n[+human, MASC]
d.n[MASC]
e.n[–human]
f.n
Both n flavors in (15a) and (15b) derive feminine nouns, but the difference between the two is that the assignment of feminine gender is semantically grounded in the former case (as, for example, in the noun mitéra, ‘mother’) but not in the latter (as, for example, in the noun platía, ‘square’). In the same way, the two n flavors in (15c) and (15d) can explain the difference between masculine nouns such as patéras, ‘father’ and pínakas, ‘painting’. Finally, the n[–human] flavor (15e) is present in the structure of inanimate nouns like máθima, ‘lesson’ and is morphologically interpreted as neuter, while the n flavor in (15f) represents the default neuter gender that is assigned in the absence of any grammatical, lexical, or pragmatic information.
Based on the above, we posit the following grammatical structure for Greek nouns:11
(16)Languages 10 00085 i003
The syntactic structure in (16) is realized by phonological exponents, namely, by phonological representations associated with specific grammatical features/feature bundles in the mental lexicon. For example, for the nominal form kiθaristón, ‘guitar players (MASC, PL, GEN)’, we could assume the syntactic structure (after head movement; see, e.g., Embick & Noyer, 2007; Embick, 2010) and the exponents presented in (17)Embick, 2010) and the exponents presented in (17)12.
(17)Languages 10 00085 i004

3.2. n Exponents

We saw in Table 2 that the nouns with the stem-final sequences /að/ and /ið/ consistently denote human referents. We also saw that any noun that includes the stem-final sequence /at/ refers to non-human objects/entities. Putting together this empirical observation and the theoretical framework we sketched out in Section 3.1, we could argue that /að/, /ið/, and /at/ are n exponents (see also Markopoulos, 2021; Markopoulos & Frantzi, 2022; cf. Markopoulos, 2018). In particular, /að/ and /ið/ can realize an n node specified as [+human, MASC], while /at/ is an exponent of the n[–human] flavor, as exemplified by the nominal forms papáðes, ‘priests’; manáviðes, ‘grocers’; and pirámata, ‘experiments’ in (18–20)13.
(18)Languages 10 00085 i005
(19)Languages 10 00085 i006
(20)Languages 10 00085 i007

3.3. Theme Exponents

Let us now move on to the other category of stem-final segments, namely, /o/, /a/, and /i/. Recall from Table 1 that these segments may appear in both human and non-human nouns of any grammatical gender. It is thus clear that they cannot be exponents of any n flavor. Moreover, they do not seem to realize any specific grammatical feature, given that their distribution is, for the most part, random. A question then arises as to what the role of such a segment in a nominal form is.
Following previous analyses of other languages (e.g., Oltra-Massuet, 1999 on Catalan; Oltra-Massuet & Arregi, 2005 on Spanish; Embick & Noyer, 2007 on Latin; Halle & Nevins, 2009 on Slavic languages) and building on Revithiadou & Spyropoulos’s (2016) analysis of Greek nominals, we propose that the stem-final segments above are exponents of a Theme node, which is inserted next to the n node post-syntactically, due to language-specific well-formedness requirements (see Section 3.1).14 This Theme node bears no grammatical features and simply satisfies the structural restrictions of Greek nominal inflection. The post-syntactic insertion of the Theme node (indicated by the dotted line) is schematically presented in (21).
(21)Languages 10 00085 i008
Based on the template in (21), we may posit the following structures for the nominal forms uranós, ‘sky (MASC)’; θálasa, ‘sea (FEM)’; and planítis, ‘planet (MASC)’. In all three cases, the n node is not overtly expressed, and the Theme node is realized by the vowels /o/, /a/, and /i/, respectively.15
(22)Languages 10 00085 i009
(23)Languages 10 00085 i010
(24)Languages 10 00085 i011
In summary, we have proposed that what has traditionally been referred to as a “nominal stem” amounts to the phonological realization of three distinct terminal nodes: √root, n, and Theme. The stem-final sequences /að/, /ið/, and /at/ are exponents of the n node, which explains why they appear in nouns with certain morphosyntactic and semantic properties. In contrast, the stem-final vowels /o/, /a/, and /i/ function as exponents of the Theme node, which carries no grammatical features. Consequently, these vowels may emerge in both human and non-human nouns of any grammatical gender.16

4. Phonological Analysis

4.1. Segmental Representation

In Section 3.2, we argued that the nominal forms papáðes (/pap-að-Ø-es/) and manáviðes (/manav-ið-Ø-es/) include the n[+human, MASC] exponents /að/ and /ið/, respectively. Similarly, we claimed that, in the form pirámata (/piram-at-Ø-a/), what precedes the PL exponent /a/ is the n[–human] exponent /at/. These assumptions give rise to an empirical problem: Why do the singular forms papás, manávis, and pírama lack the consonants /ð/ and /t/?
We hypothesize that the answer lies in the phonological computation of these forms (see also Markopoulos, 2018, 2021; Markopoulos & Frantzi, 2022). More specifically, we propose that syntax does deliver to phonology the sequences /pap-að-Ø-s/, /manav-ið-Ø-s/, and /piram-at-Ø-Ø/; however, the stem-final consonants /ð/ and /t/ are not eventually realized. This is because these consonants are not “strong” enough to surface in all phonological environments. Unlike the other segments of the three nominal forms, which are always pronounced, /ð/ and /t/ are weaker and emerge only when the phonological environment allows so.
To formalize the above hypothesis, we employ the framework of GHG (Smolensky & Goldrick, 2016; Rosen, 2016; Faust & Smolensky, 2017; Zimmermann, 2018, 2021; Hsu, 2022).17 GHG is a constraint-based phonological model according to which every phonological element is specified with an Activity Level (AL) value that expresses its underlying strength. This value is greater than 0 and less than or equal to 1. Any element with an AL lower than 1 needs to be reinforced in order to reach the realization threshold (=1) and be pronounced (cf. Zimmermann, 2018; Markopoulos & Apostolopoulou, 2022). For instance, a consonant with AL equal to 0.7 will be realized only if it receives an extra 0.3 of activity from the phonological grammar.
Reinforcing a weak element with epenthetic activity entails the violation of the constraint Dep, which requires that the amount of activity in the output must correspond to the same amount of activity in the input (Smolensky & Goldrick, 2016). On the other hand, the non-realization of an underlying element triggers the violation of the constraint Max, which necessitates the presence of any amount of input activity in the output (Faust & Smolensky, 2017). The significance of each violation is determined not only by the underlying AL of the element but also by the weight of the relevant constraint. For example, in the toy tableau in (25), the phonological input includes a weak consonant /t/ with AL 0.7 (t0.7).18 The candidate in (25a) violates Dep, because the realization of this consonant requires 0.3 of epenthetic activity. The penalty for this violation equals the amount of epenthetic activity multiplied by the weight of the constraint. On the contrary, the candidate in (25b) leaves /t0.7/ unrealized and, therefore, incurs a violation of Max, which receives a penalty equal to the amount of unrealized activity (given that, in this case, the weight of the constraint is 1).
(25) /at0.7/Dep
weight: 2
Max
weight: 1
a.[at1]−(1 − 0.7) × 2 = −0.6
b.[a] −(0.7 × 1) = −0.7
Of course, the constraint system that evaluates the possible outputs usually consists of more than two constraints. The well-formedness of an output could thus be measured by adding up the penalty scores for each constraint. The candidate with the lowest total penalty score, or, in GHG terms, with the highest Harmony (H) value, would be the optimal one.
For example, the evaluation of the outputs in (25) could include a constraint that disallows word-final stop obstruents (*C[–son, –cont]]ω), the violation of which would be penalized with a negative number equal to the constraint’s weight. In the tableau in (26), the output [at1] (26a) receives two penalties for the violation of Dep and C]ω, which yields the H value −1.6. On the other hand, the candidate in (26b) has a higher H value, because it violates only Max and, therefore, is optimal in this case.
(26) /at0.7/Dep
weight: 2
Max
weight: 1
*C[–son, –cont]]ω
weight: 1
H
a.[at1]19−(1 − 0.7) × 2 = −0.6 −1−1.6
b.[a] −(0.7 × 1) = −0.7 −0.7
Let us now return to the n exponents /að/, /ið/, and /at/ and to the reason why the consonants of these exponents do not appear in the singular forms papás, manávis, and pírama. To begin with /at/, we assume that the consonant /t/ of the exponent is weak, with an AL equal to 0.7.20 Consequently, the phonological computation of the exponent /at/ given the input /piram-at0.7/ is as presented in the tableau in (26). The output [ˈpirama] is preferred over the output [ˈpiramat] because it has the highest H value.
On the contrary, in the plural form pirámata (/piram-at0.7-a), /at0.7/ is followed by the vocalic exponent /a/, which creates a more suitable phonological environment for /t0.7/ to be realized. First, the constraint *C]ω is no longer pertinent. What is more, the output [piˈramaa] (27b) violates the constraint *VV, which prohibits the occurrence of hiatus, whereas [piˈramata] (27a) does not. As a result, candidate (27a) is preferable.
(27) /piram-at0.7-a/Dep
weight: 2
Max
weight: 1
*VV
weight: 1
H
a.[piˈramata]−(1 − 0.7) × 2 = −0.6 −0.6
b.[piˈramaa] −(0.7 × 1) = −0.7−1−1.7
In the same vein, we propose that the consonant /ð/ of the exponents /að/ and /ið/ is weak (/ð0.7/) and is silenced in singular forms (e.g., /pap-að0.7-s/ → [paˈpas]; /manav-ið0.7-s/ → [maˈnavis]), not only due to the required amount of epenthetic activity that violates Dep, but also due to the violation of the constraint *CC]ω, which disallows word-final clusters. The computation of the input /pap-að-s/ is presented in (28).
(28) /pap-að0.7-s/Dep
weight: 2
Max
weight: 1
*CC]ω
weight: 2
H
a.[paˈpaðs]−(1 − 0.7) × 2 = −0.6 −2−2.6
b.[paˈpas] −(0.7 × 1) = −0.7 −0.7
The picture changes in plural forms, where the exponent /að0.7/ is followed by the PL exponent /es/. In this case, not only is there no violation of *CC]ω, but also the omission of /ð0.7/ results in a hiatus penalized by *VV. As a result, the output that eventually surfaces is [paˈpaðes], as illustrated in (29).
(29) /pap-að0.7-es/Dep
weight: 2
Max
weight: 1
*VV
weight: 1
H
a.[paˈpaðes]−(1 − 0.7) × 2 = −0.6 −0.6
b.[paˈpaes] −(0.7 × 1) = −0.7−1−1.7
A similar analysis is proposed for the Theme exponents /o/, /a/, and /i/. These exponents have an AL equal to 0.7 and are realized only in singular forms, owing to the favorable phonological environment (e.g., /uran-o0.7-s/ → [uraˈnos]; /θalas-a0.7/ → [ˈθalasa]; /planit-i0.7-s/ → [plaˈnitis]). Conversely, they fail to surface in plural forms, as their presence would result in hiatus (e.g., /uran-o0.7-i/ → [uraˈni]; /θalas-a0.7-es/ → [ˈθalases]; /planit-i0.7-es/ → [plaˈnites]). The computation given the inputs /θalas-a/ and /θalas-a-es/ is presented in detail in (30–31).
(30) /θalas-a0.7/Dep
weight: 2
Max
weight: 1
*VV
weight: 1
H
a.[ˈθalasa]−(1 − 0.7) × 2 = −0.6 −0.6
b.[ˈθalas] −(0.7 × 1) = −0.7 −0.7
(31) /θalas-a0.7-es/Dep
weight: 2
Max
weight: 1
*VV
weight: 1
H
a.[θaˈlasaes]−(1 − 0.7) × 2 = −0.6 −1−1.6
b.[ˈθalases] −(0.7 × 1) = −0.7 −0.7

4.2. Stress

In Section 2.1, we saw that nouns with the n exponents /að/, /ið/, and /at/ have predictable stress patterns: nouns with /að/ are always stressed on the vowel of the n exponent (e.g., pap-áð-es), nouns with /ið/ are always stressed on the syllable that precedes the exponent (e.g., manáv-ið-es), and nouns with /at/ are always stressed two syllables to the left of the exponent (e.g., pirám-at-a). In other words, /að/ always receives stress, while /ið/ and /at/ require stress to be placed one or two syllables to the left, respectively. We thus conclude that these exponents seem to inherently carry information about the stress position of the nominal forms they are part of.
Drawing on Revithiadou’s (1999, 2007) hypothesis that Greek morphemes bear inherent lexical accents and on Revithiadou & Spyropoulos’ (2016) analysis of the stress properties of Greek theme elements, we posit that the underlying representation of the n exponents /að/, /ið/, and /at/ includes a trochaic foot (cf. Markopoulos, 2018). In the case of /að/, the vowel of the exponent is associated with the head of the trochee, while in /ið/ it is associated with its tail. Finally, /at/ requires that a trochaic foot be associated with the two syllables on its left. Thus, the underlying representations of the three exponents are formed as follows.
(32)(*   .)
0.7 - σ
(33)(*  .)
σ - ið0.7
(34)(* .)
σ σ - at0.7
Given that these lexical accents are almost always realized, we may assume that they have an AL equal or very close to 1. On the other hand, Theme exponents lack such strong underlying accents, which is another key factor distinguishing them from n exponents. However, this does not mean that Theme exponents have no underlying accents whatsoever. Previous studies (Apostolouda, 2012, 2018; Revithiadou & Lengeris, 2016) have shown a clear correlation between Theme exponents like /o/, /a/, and /i/ and specific stress patterns. This correlation is not only found in the lexicon (i.e., in lexical databases) but is also confirmed in experimental tasks, where native speakers tend to assign stress to pseudo-nouns based on the specific Theme exponent in each case.
Apostolouda’s (2018) research on two lexical databases (Annotated-Lexicon (ALex) and Tagged-Clean (TC)) revealed that the distribution of stress patterns in the lexicon is not uniform across noun classes.21 For instance, if we compare three-syllable masculine nouns with the Theme exponent /o/ and three-syllable masculine nouns with the Theme exponent /i/, we observe a clear difference in the percentage of APU and PU stress. The frequencies for each stress pattern per noun class are presented in Table 4 (Apostolouda, 2018, pp. 171, 173).
Interestingly, the distribution of stress patterns in the lexicon was largely mirrored in the experimental results. For example, when asked to choose among oxytone, paroxytone, and proparoxytone versions of pseudo-nouns ending in -os (-ος) and -is (-ης), i.e., pseudo-nouns expected to be interpreted as masculine nouns with the Theme exponents /o/ and /i/, adult Greek speakers showed a preference for APU stress in nouns ending in -os and a preference for PU stress in nouns ending in -is, as illustrated in Table 5 (Apostolouda, 2018, pp. 144–145).
The findings above suggest that Theme exponents, just like n exponents, do have inherent stress properties, albeit not strong enough so that their lexical accent is realized in all environments.22 This provides further evidence in favor of the claim that stem endings are separate morphological units with their own grammatical profiles. The only difference between the two types of exponents, as far as stress is concerned, is the strength level of their underlying lexical accents.

5. Discussion

In the previous sections, we argued that the stem endings of Greek nouns are characterized by distinct grammatical properties and, therefore, should be treated as independent morphological items. Furthermore, we claimed that the allomorphic variation exhibited by these stem endings is due to their gradient underlying representation, which prevents them from being faithfully pronounced in all phonological environments.
In an alternative approach, the segments/sequences -a(ð), -i(ð), -a(t), -o, -a, and -i could be analyzed as part of the nominal stem. Under this view, the allomorphy patterns -á/-áð, -i/-, -a/-at, -o/-Ø, -a/-Ø, and -i/-Ø could be accounted for by postulating different stem allomorphs for each noun (see Booij, 1997; Ralli, 2000, 2006, 2022; Karasimos, 2011; Anastassiadis-Symeonidis, 2012; Marinis, 2019). For example, for the nouns papás, manávis, and pírama, which exhibit the allomorphy pattern -V/-VC, one could postulate the following stem allomorphs.
(35)a.papa-~papað
b.manavi-~manavið
c.pirama-~piramat-
Each allomorph in (35) is associated with specific morphological environments. The allomorphs in the first column are used in singular forms, whereas those in the second column are typically found in plural forms.
A similar analysis could be proposed for the nouns uranós, θálasa, and planítis, which exhibit the allomorphy pattern -V/-Ø.
(36)a.urano-~uran
b.θalasa-~θalas
c.planiti-~planit-
Such an approach may appear simple and straightforward, but it fails to capture important empirical generalizations. First, it disregards the systematic correlation between stem endings and specific morphosyntactic and semantic features. For instance, under this analysis, the fact that all nouns exhibiting the allomorphy pattern -i/- are masculine and refer to humans would be coincidental. Second, it overlooks the fact that the stress pattern of a noun is largely dependent on its stem ending. Third, it does not highlight the phonological aspects of the allomorphic phenomena under investigation. These phenomena may not be textbook instances of phonologically conditioned allomorphy, but, nonetheless, they have a clear phonological motivation: the avoidance of ill-formed phonological outputs such as hiatus, word–final clusters, etc.
The constituenthood of the n exponents under investigation is further supported by their emergence in novel environments. Neologisms such as tip-áð-es, ‘guys’, and rezíl-ið-es, ‘deplorable people’, indicate that the exponents /að/ and /ið/ are not mere endings of existent stems, but they can be employed in the formation of new masculine human nouns. Another piece of evidence comes from the tendency of several masculine human nouns with the stem-final /a/ or /i/ in singular forms to develop alternative plural forms with the exponent /að/ or /ið/ (e.g., patér-a-s ‘father (SG)’, cf. PL: patér-es/pater-áð-es; vivli-o-pól-i-s ‘bookseller (SG)’, cf. PL: vivli-o-pól-es/vivli-o-pól-ið-es). This tendency underscores the strong association between the exponents /að, ið/ and the feature bundle [+human, MASC].
What is more, the distinction between n and Theme exponents proposed in the present analysis captures the fact that only n exponents are expected to appear in new environments. Apart from the neologisms presented above, the exponents /að/ and /ið/ are also used for the adaptation of loanwords referring to humans (e.g., deliver-áð-es, ‘delivery men’ (<English delivery), tsíp-ið-es, ‘stingy people’ (<English cheap)). On the other hand, Theme exponents are rarely encountered in new word formations, which suggests that they are lexically specified to combine only with existent root exponents and are no longer productive.23
Capturing the difference between n and Theme exponents necessitates a framework that allows post-syntactic operations. As discussed in Section 3, DM enables the distinction between, on the one hand, exponents of nominalizing heads bearing morphosyntactic and semantic features, and on the other hand, exponents of Theme nodes, which are inserted post-syntactically and have no morphosyntactic or semantic content. In other words, unlike nominalizers, Theme nodes do not seem to have any syntactically meaningful function in the Greek nominal structure; therefore, they should be inserted after the syntactic derivation is complete. Previous studies focusing on verbal morphology have often associated Theme vowels with specific morphosyntactic features/functions, e.g., voice/aspect (Warburton, 1973; Galani, 2003, 2005), argument structure (Jabłońska, 2007; Taraldsen Medová & Wiland, 2019), verbalizers/light verbs (Kayne, 2016; Fábregas, 2017); see also Kovačević et al. (2022) for a detailed discussion. In this case, however, the vocalic exponents of Theme nodes cannot be assigned any specific syntactic role, given that, as we saw in Section 2, their distribution is not systematic.24

6. Conclusions

The primary claim of this article is that Greek nominal stems break down into separate constituents. To support this claim, we provided evidence drawn both from morphosyntax and phonology. More specifically, we demonstrated that the stem-final sequences /að/, /ið/, and /at/ assign morphosyntactic (i.e., grammatical gender) and semantic (i.e., [±human]) properties to nouns and, additionally, determine the position of stress. On the other hand, the stem-final segments /o/, /a/, and /i/ do not convey any specific morphosyntactic or semantic property, yet, importantly, they can influence the probability of a stress pattern to emerge. These empirical findings suggest that all six stem-final segments/sequences in question are independent morphological units and fall into two categories: /að/, /ið/, and /at/ are exponents of the n head that assigns to roots the grammatical category of the noun, whereas /o/, /a/, and /i/ are exponents of the Theme node, which bears no grammatical features and is inserted due to language-specific restrictions.
The proposed analysis not only highlights the independent status and the specific properties of stem-final segments/sequences, which have been overlooked in the relevant literature, but it also draws a distinction between stem endings that realize nominalizing heads and stem endings that function as Theme exponents. Thus, it captures the fact that only the former can define the morphosyntactic profile of a noun and can appear in new word formations. Finally, it offers new insights into the study of Greek nominal inflection by allowing for an organization of the nominal system that goes beyond the traditional tripartite gender distinction (FEM/MASC/NEUT), crucially taking into account the semantic (i.e., [±human]) and the phonological (i.e., stress pattern) properties of each noun category.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

No new data were created or analyzed in this study. Data sharing is not applicable to this article.

Acknowledgments

I would like to thank two anonymous reviewers for their insightful remarks, as well as Anthi Revithiadou, Vassilios Spyropoulos, and Eirini Apostolopoulou for discussion and comments on previous versions of this article.

Conflicts of Interest

The author declares no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
ACCAccusative
ALActivity Level
ALexAnnotated-Lexicon
APUAntepenultimate
CConsonant
DMDistributed Morphology
FEMFeminine
GENGenitive
GHGGradient Harmonic Grammar
HHarmony
MASCMasculine
nNominalizer
NEUTNeuter
NOMNominative
NumNumber
PLPlural
PUPenultimate
REFReferentiality
SGSingular
TCTagged-Clean
UUltimate
VVowel
VOCVocative

Notes

1
We define as “stem” the part of the nominal form that results if we remove any number/case exponents (cf. Anastassiadis-Symeonidis, 2012; Ralli, 2022).
2
Here, we exclude phenomena of a phonological nature that may be attested at the left edge of the nominal form, as, for example, assimilation of stops after nasals (e.g., /tin porta/ → [timˈboɾta], ‘the door’; see Revithiadou & Markopoulos, 2021).
3
All neuter and most feminine nouns have a syncretic nominative-accusative-vocative form. To ease the readability of the examples, we omit the vocative case in glosses.
4
The term permissible echoes the trisyllabic window restriction in Greek, i.e., the requirement that stress may fall on only one of the last three syllables of the word (Malikouti-Drachman & Drachman, 1989; Drachman & Malikouti-Drachman, 1999; Revithiadou, 1999, 2007).
5
There are also neuter nouns that, at the surface level, have a stem ending in -i (e.g., spíti-Ø ‘house.NEUT-SG.NOM/ACC’; psomí-Ø ‘bread.NEUT-SG.NOM/ACC’). However, the different allomorphy pattern that is attested in these nouns (e.g., /spitj-a/ [ˈspitça] ‘house.NEUT-PL.NOM/ACC’; /psomj-a/ [psoˈmɲa] ‘bread.NEUT-PL.NOM/ACC’) suggests that the surface stem-final /i/ in fact corresponds to an underlying /j/ (see Markopoulos, 2018; Apostolopoulou, 2018).
6
The pattern -i/- emerges exclusively in masculine human nouns. The pattern -á/-áð is also attested in a handful of feminine nouns (e.g., mamá ‘mom (SG)’, cf. mamáð-es (PL)) and in a closed class of Turkish non-human loanwords (e.g., kuvá-s ‘bucket (SG)’, cf. kuváð-es (PL)).
7
Νouns with stem-final /i(ð)/ that include the derivational suffixes -dz and -l are always stressed on the last syllable of the stem (e.g., taksi-dz-í-s ‘taxi driver (SG)’, cf. taksi-dz-íð-es (PL); merak-l-í-s ‘a person keen on something (SG)’, cf. merak-l-íð-es (PL)). Here we consider only non-derived nouns; therefore, this noun category is excluded.
8
When the singular form consists of two syllables, it receives stress on the leftmost one, i.e., on PU (e.g., kíma ‘wave (SG)’).
9
Apart from DM, the decomposition of a stem into separate constituents could also be formalized within other syntax-oriented frameworks such as the Exo-Skeletal Model (Borer, 2005a, 2005b, 2013), Spanning (Svenonius, 2012), Nanosyntax (Starke, 2009), or Morphology as Syntax (Collins & Kayne, 2023). Of course, the implementation would require framework-specific adjustments in each case. For instance, in a nanosyntactic account, the root and n nodes would break down into a sequence of functional heads. To give a brief example, the structure of a feminine nominal stem could be the following: [FEM [CLASS REF]] (see Caha, 2025, to appear). I would like to thank an anonymous reviewer for bringing up this issue.
10
See Markopoulos (2018) and Alexiadou & Anagnostopoulou (2023) for further examples and discussion.
11
In Greek, the n node has also been argued to host (post-syntactic) declension class features (Alexiadou, 2017; Alexiadou & Anagnostopoulou, 2023; see also Kramer, 2015), as well as Number features in idiosyncratic plural forms of, e.g., mass nouns, such as ner-á, ‘water-PL’ (Alexiadou, 2011; see also Lecarme, 2002; Acquaviva, 2008; Lowenstamm, 2008; Kramer, 2016; Punske & Jackson, 2017). Many thanks to an anonymous reviewer for raising this point.
12
Here we remain agnostic about the (non-)content of roots. To enhance the clarity of the examples, in the remainder of the article, we refer to roots using abstract meanings written in small caps. This choice, however, should not be interpreted as a particular approach to the nature of roots. For some discussion, see, among others, Arad (2005); Acquaviva (2009); Borer (2013); Alexiadou et al. (2014); Harley (2014); Panagiotidis (2014, 2020).
13
Given that the inflectional suffix -es in pap-áð-es and manáv-ið-es and the suffix -a in pirám-at-a are underspecified for case (they may appear in nominative, accusative, or vocative plural forms), they are taken to be elsewhere PL exponents.
14
See also Thomadaki (1994) for a lexicalist approach to Theme vowels in Greek.
15
The Theme node is also present in the syntactic structures in (18–20) but has a zero exponent.
16
Revithiadou and Spyropoulos (2016) also highlight the distinction between, on the one hand, /o/, /a/, and /i/, and, on the other hand, /að/, /ið/, and /at/, but propose that both categories of stem endings are Theme exponents. The Theme node that is realized by the former category does not attach to the n node, but to a functional category F higher in the structure.
17
Alternatively, this hypothesis could be framed within an analysis positing floating segments (e.g., Faust, 2014; Scheer, 2016).
18
Throughout this section, a subscript number is used to indicate the AL of an element.
19
In theory, in each computation there is an infinite number of possible outputs. In the present analysis, we focus only on the top two candidates, one where the weak segment is realized and one where it is not pronounced.
20
In the present study, AL values are assigned arbitrarily and simply serve as indicators of the elements’ relative strength. For some approaches to calculating AL using lexical databases, see Markopoulos & Frantzi (2022) and Markopoulos et al. (2024).
21
ALex is based on Anastassiadis-Symeonidis’ (2002Reverse Dictionary of Modern Greek and consists of 5,133 nouns (types). TC is based on Protopapas et al.’s (2012) Clean Corpus and includes 22,806 nominal forms (types and tokens) (Apostolouda, 2018, p. 164).
22
Within the framework of GHG, this would translate into AL lower than 1. A detailed GHG account of stress assignment is beyond the scope of the present article. For an analysis that captures the probabilistic nature of stress assignment in Greek nouns, see Markopoulos et al. (2024).
23
See Christofidou (2003) and Anastassiadis-Symeonidis (2012) for further evidence and discussion.
24
I am grateful to an anonymous reviewer for bringing up this discussion.

References

  1. Acquaviva, P. (2008). Lexical plurals: A morphosemantic approach. Oxford University Press. [Google Scholar]
  2. Acquaviva, P. (2009). Roots and lexicality in Distributed Morphology. York Papers in Linguistics (Series 2), 10, 1–21. [Google Scholar]
  3. Alexiadou, A. (2011). Plural mass nouns and the morpho-syntax of number. In M. B. Washburn, K. McKinney-Bock, E. Varis, A. Sawyer, & B. Tomaszewicz (Eds.), Proceedings of the 28th west coast conference on formal linguistics (pp. 33–41). Cascadilla Proceedings Project. [Google Scholar]
  4. Alexiadou, A. (2017). Gender and nominal ellipsis. In N. LaCara, K. Moulton, & A.-M. Tessier (Eds.), “A schrift to fest kyle johnson” (Vol. 1, pp. 11–22). Linguistics Open Access Publications. [Google Scholar]
  5. Alexiadou, A., & Anagnostopoulou, E. (2023). Zero-derived nouns in Greek. Languages, 8, 13. [Google Scholar] [CrossRef]
  6. Alexiadou, A., Borer, H., & Schäfer, F. (2014). The syntax of roots and the roots of syntax. Oxford University Press. [Google Scholar]
  7. Alexiadou, A., Haegeman, L., & Stavrou, M. (2007). Noun phrase in the generative perspective. Mouton de Gruyter. [Google Scholar]
  8. Anastassiadis-Symeonidis, A. (2002). Aντίστροφο λεξικό της Νέας Ελληνικής [Reverse dictionary of Modern Greek]. Institute of Modern Greek Studies. [Google Scholar]
  9. Anastassiadis-Symeonidis, A. (2012). Το νεοελληνικό κλιτικό σύστημα των ουσιαστικών και οι τάσεις του [The Modern Greek noun inflectional system and its tendencies]. In Z. Gavriilidou, A. Efthymiou, E. Thomadaki, & P. Kambakis-Vougiouklis (Eds.), Selected papers of the 10th international conference of Greek linguistics (pp. 23–40). Democritus University of Thrace. [Google Scholar]
  10. Anastassiadis-Symeonidis, A., & Cheila-Markopoulou, D. (2003). Συγχρονικές και διαχρονικές τάσεις στο γένος της Ελληνικής (Μια θεωρητική πρόταση) [Synchronic and diachronic tendencies in Greek gender (A theoretical proposal)]. In A. Anastassiadis-Symeonidis, A. Ralli, & D. Cheila-Markopoulou (Eds.), Το γένος [Gender] (pp. 13–56). Patakis. [Google Scholar]
  11. Apostolopoulou, E. (2018). H φωνολογική αναπαράσταση του ημιφώνου της Κοινής Νέας Ελληνικής: Μια προσέγγιση με βάση τη Διαβαθμισμένη Aρμονική Γραμματική [The phonological representation of glide in Standard Modern Greek: An approach based on Gradient Harmonic Grammar] [Master’s Thesis, Aristotle University of Thessaloniki]. [Google Scholar]
  12. Apostolouda, V. (2012). O τονισμός των ουσιαστικών της Ελληνικής: Μια πειραματική προσέγγιση [The stress of Greek nouns: An experimental approach] [Master’s Thesis, Aristotle University of Thessaloniki]. [Google Scholar]
  13. Apostolouda, V. (2018). Πειραματικές διερευνήσεις στον τονισμό της Ελληνικής [Experimental investigations on Greek stress] [Ph.D. Thesis, Aristotle University of Thessaloniki]. [Google Scholar]
  14. Arad, M. (2005). Roots and patterns: Hebrew morpho-syntax. Springer. [Google Scholar]
  15. Booij, G. (1997). Allomorphy and the autonomy of morphology. Folia Linguistica, 31(1–2), 25–56. [Google Scholar] [CrossRef]
  16. Borer, H. (2005a). In name only: Structuring sense (Vol. I). Oxford University Press. [Google Scholar]
  17. Borer, H. (2005b). The normal course of events: Structuring sense (Vol. II). Oxford University Press. [Google Scholar]
  18. Borer, H. (2013). Taking form: Structuring sense (Vol. III). Oxford University Press. [Google Scholar]
  19. Caha, P. (2025). Root and stem allomorphy without multiple exponence: The case of special nominatives. Morphology. Published online 2024. [Google Scholar] [CrossRef]
  20. Caha, P. (to appear). Nanosyntax: Some key features. In A. Alexiadou, R. Kramer, A. Marantz, & I. Oltra-Massuet (Eds.), The Cambridge handbook of distributed morphology. Cambridge University Press. [Google Scholar]
  21. Christofidou, A. (2003). Γένος και κλίση στην Ελληνική (Μια φυσική προσέγγιση) [Gender and inflection in Greek (A natural approach)]. In A. Anastassiadis-Symeonidis, A. Ralli, & D. Cheila-Markopoulou (Eds.), Το γένος [Gender] (pp. 100–131). Patakis. [Google Scholar]
  22. Clairis, C., & Babiniotis, G. (2004). Γραμματική της Νέας Ελληνικής: Δομολειτουργική-επικοινωνιακή [Grammar of modern Greek: Structural-functional-communicative]. Ellinika Grammata. [Google Scholar]
  23. Collins, C., & Kayne, R. S. (2023). Towards a theory of morphology as syntax. Studies in Chinese Linguistics, 44(1), 1–32. [Google Scholar] [CrossRef]
  24. Drachman, G., & Malikouti-Drachman, A. (1999). Greek word accent. In H. van der Hulst (Ed.), Word prosodic systems in the languages of Europe (pp. 897–945). Mouton de Gruyter. [Google Scholar]
  25. Embick, D. (2010). Localism versus globalism in morphology and phonology. MIT Press. [Google Scholar]
  26. Embick, D., & Noyer, R. (2007). Distributed Morphology and the syntax-morphology interface. In G. Ramchand, & C. Reiss (Eds.), The Oxford handbook of linguistic interfaces (pp. 289–324). Oxford University Press. [Google Scholar]
  27. Faust, N. (2014). Where it’s [at]: A phonological effect on phasal boundaries in the construct state of Modern Hebrew. Lingua, 150, 315–331. [Google Scholar] [CrossRef]
  28. Faust, N., & Smolensky, P. (2017). Activity as an alternative to autosegmental association [Unpublished manuscript]. Université Paris 8 CNRS/SFL & Johns Hopkins University. [Google Scholar]
  29. Fábregas, A. (2017). Theme vowels are verbs. Rivista di Grammatica Generativa, 39, 1–11. [Google Scholar]
  30. Galani, A. (2003). An analysis of theme vowels in Modern Greek within Distributed Morphology. Interlingüística, 14, 399–412. [Google Scholar]
  31. Galani, A. (2005). The morphosyntax of verbs in Modern Greek [Ph.D. Thesis, University of York]. [Google Scholar]
  32. Halle, M., & Marantz, A. (1993). Distributed Morphology and the pieces of inflection. In K. Hale, & S. J. Keyser (Eds.), The view from building 20: Essays in linguistics in honor of Sylvain Bromberger (pp. 111–176). MIT Press. [Google Scholar]
  33. Halle, M., & Marantz, A. (1994). Some key features of Distributed Morphology. In A. Carnie, & H. Harley (Eds.), MIT working papers in linguistics 21: Papers on phonology and morphology (pp. 275–288). MITWPL. [Google Scholar]
  34. Halle, M., & Nevins, A. (2009). Rule application in phonology. In E. Raimy, & C. E. Cairns (Eds.), Contemporary views on architecture and representations in phonology (pp. 355–382). MIT Press. [Google Scholar]
  35. Harley, H. (2014). On the identity of roots. Theoretical Linguistics, 40(3–4), 225–276. [Google Scholar] [CrossRef]
  36. Harley, H., & Noyer, R. (1999). Distributed Morphology. Glot International, 4(4), 3–9. [Google Scholar]
  37. Holton, D., Mackridge, P., Philippaki-Warburton, I., & Spyropoulos, V. (2012). Greek: A comprehensive grammar (2nd ed.). Routledge. [Google Scholar]
  38. Hsu, B. (2022). Gradient symbolic representations in Harmonic Grammar. Language and Linguistics Compass, 16(9), e12473. [Google Scholar] [CrossRef]
  39. Jabłońska, P. (2007). Radical decomposition and argument structure [Ph.D. Thesis, University of Tromsø]. [Google Scholar]
  40. Karasimos, A. (2011). Υπολογιστική επεξεργασία της αλλομορφίας στην παραγωγή λέξεων της Ελληνικής [Computational processing of allomorphy in Greek word derivation] [Ph.D. Thesis, University of Patras]. [Google Scholar]
  41. Kayne, R. S. (2016). What is suppletive allomorphy? On went and on *goed in English [Unpublished manuscript]. New York University. [Google Scholar]
  42. Kihm, A. (2005). Noun class, gender, and the lexicon-syntax-morphology interfaces. In G. Cinque, & R. S. Kayne (Eds.), The Oxford handbook of comparative syntax (pp. 459–512). Oxford University Press. [Google Scholar]
  43. Kovačević, P., Antonyuk, S., & Quaglia, S. (2022). The status of verbal theme vowels in contemporary linguistic theorizing: Some recent developments. An introduction. Glossa, 9(1), 1–41. [Google Scholar] [CrossRef]
  44. Kramer, R. (2015). The morphosyntax of gender. Oxford University Press. [Google Scholar]
  45. Kramer, R. (2016). A split analysis of plurality: Number in Amharic. Linguistic Inquiry, 47(3), 527–559. [Google Scholar] [CrossRef]
  46. Lecarme, J. (2002). Gender “polarity”: Theoretical aspects of Somali nominal morphology. In P. Boucher, & M. Plénat (Eds.), Many morphologies (pp. 109–141). Cascadilla Press. [Google Scholar]
  47. Lowenstamm, J. (2008). On little n, √, and types of nouns. In J. Hartmann, V. Hegedus, & H. van Riemsdijk (Eds.), Sounds of silence: Empty elements in syntax and phonology (pp. 105–144). Elsevier. [Google Scholar]
  48. Malikouti-Drachman, A., & Drachman, G. (1989). Τονισμός στα Ελληνικά [Stress in Greek]. Studies in Greek Linguistics, 9, 127–143. [Google Scholar]
  49. Marinis, M. (2019). Κλίση και οργάνωση κλιτικών παραδειγμάτων [Inflection and organization of inflectional paradigms] [Ph.D. Thesis, University of Patras]. [Google Scholar]
  50. Markopoulos, G. (2018). Phonological realization of morphosyntactic features [Ph.D. Thesis, Aristotle University of Thessaloniki]. [Google Scholar]
  51. Markopoulos, G. (2021). Nominal inflection in the morphosyntax-phonology interface: A comparative study of Greek and Hebrew. In T. Markopoulos, C. Vlachos, A. Archakis, D. Papazachariou, A. Roussou, & G. J. Xydopoulos (Eds.), Proceedings of the ICGL14 (pp. 792–801). University of Patras. [Google Scholar]
  52. Markopoulos, G. (2023). Το ονοματικό σύστημα της Ελληνικής [The nominal system of Greek]. In D. Papadopoulou, & A. Revithiadou (Eds.), Εισαγωγή στη μορφολογία [Introduction to morphology] (pp. 116–149). Institute of Modern Greek Studies. [Google Scholar]
  53. Markopoulos, G., & Apostolopoulou, E. (2022). Vowel height as sonority level. In A.-M. Sougari, & V. Bardzokas (Eds.), Selected papers on theoretical and applied linguistics from ISTAL 24 (pp. 513–529). Aristotle University of Thessaloniki. [Google Scholar]
  54. Markopoulos, G., & Frantzi, K. (2022). Investigating stem allomorphy in Greek and Arabic by means of lexical frequency. In A.-M. Sougari, & V. Bardzokas (Eds.), Selected papers on theoretical and applied linguistics from ISTAL 24 (pp. 530–545). Aristotle University of Thessaloniki. [Google Scholar]
  55. Markopoulos, G., Apostolopoulou, E., Apostolouda, V., & Revithiadou, A. (2024). Beyond unpredictability: A GHG analysis of Greek noun stress. In S. Phadnis, C. Spellerberg, & B. Wilkinson (Eds.), Proceedings of the fifty-fourth annual meeting of the North East Linguistic Society (NELS 54) (Vol. 2, pp. 61–74). GLSA. [Google Scholar]
  56. Oltra-Massuet, I. (1999). On the notion of theme vowel: A new approach to Catalan verbal morphology [Master’s Thesis, MIT]. [Google Scholar]
  57. Oltra-Massuet, I., & Arregi, K. (2005). Stress-by-structure in Spanish. Linguistic Inquiry, 36(1), 43–84. [Google Scholar] [CrossRef]
  58. Panagiotidis, P. (2014). Indices, domains and homophonous forms. Theoretical Linguistics, 40(3–4), 415–427. [Google Scholar] [CrossRef]
  59. Panagiotidis, P. (2018). (Grammatical) gender troubles and the gender of pronouns. In É. Mathieu, M. Dali, & G. Zareikar (Eds.), Gender and noun classification (pp. 186–200). Oxford University Press. [Google Scholar]
  60. Panagiotidis, P. (2020). On the nature of roots: Content, form, identification. Evolutionary Linguistic Theory, 2(1), 56–83. [Google Scholar] [CrossRef]
  61. Protopapas, A., Tzakosta, M., Chalamandaris, A., & Tsiakoulis, P. (2012). IPLR: An online resource for Greek word-level and sublexical information. Language Resources and Evaluation, 46, 449–559. [Google Scholar] [CrossRef]
  62. Punske, J., & Jackson, S. R. (2017). The bifurcated nature of plural: Reconsidering evidence from English compounds. In J. Nee, M. Cychosz, D. Hayes, T. Lau, & E. Remirez (Eds.), Proceedings of BLS 43 (pp. 261–283). Berkeley Linguistics Society. [Google Scholar]
  63. Ralli, A. (2000). A feature-based analysis of Greek nominal inflection. Glossologia, 11–12, 201–227. [Google Scholar]
  64. Ralli, A. (2002). The role of morphology in gender determination: Evidence from Modern Greek. Linguistics, 40(3), 519–551. [Google Scholar] [CrossRef]
  65. Ralli, A. (2006). On the role of allomorphy in inflectional morphology: Evidence from dialectal variation. In G. Sica (Ed.), Open problems in linguistics and lexicography (pp. 123–152). Polimetrica. [Google Scholar]
  66. Ralli, A. (2022). Μορφολογία [Morphology] (New revised edition). Patakis. [Google Scholar]
  67. Revithiadou, A. (1999). Headmost accent wins: Head dominance and ideal prosodic form in lexical accent systems (LOT Dissertation Series 15) [Ph.D. Thesis, HIL/Leiden University]. [Google Scholar]
  68. Revithiadou, A. (2007). Colored turbid accents and containment: A case study from lexical stress. In S. Blaho, P. Bye, & M. Krämer (Eds.), Freedom of analysis? (pp. 149–174) Mouton de Gruyter. [Google Scholar]
  69. Revithiadou, A., & Lengeris, A. (2016). One or many? In search of the default stress in Greek. In J. Heinz, R. Goedemans, & H. van der Hulst (Eds.), Dimensions of phonological stress (pp. 263–290). Cambridge University Press. [Google Scholar]
  70. Revithiadou, A., & Markopoulos, G. (2021). A Gradient Harmonic Grammar account of nasals in extended phonological words. Catalan Journal of Linguistics, 20, 57–75. [Google Scholar] [CrossRef]
  71. Revithiadou, A., & Spyropoulos, V. (2016). Stress at the interface: Phases, accents and dominance. Linguistic Analysis, 4(1–2), 3–74. [Google Scholar]
  72. Rosen, E. (2016). Predicting the unpredictable: Capturing the apparent semi-regularity of rendaku voicing in Japanese through harmonic grammar. In E. Clem, V. Dawson, A. Shen, A. H. Skilton, G. Bacon, A. Cheng, & E. H. Maier (Eds.), Proceedings of BLS 42 (pp. 235–249). Berkeley Linguistics Society. [Google Scholar]
  73. Scheer, T. (2016). Melody-free syntax and phonologically conditioned allomorphy. Morphology, 26, 341–378. [Google Scholar] [CrossRef]
  74. Siddiqi, D. (2010). Distributed Morphology. Language and Linguistics Compass, 4(7), 524–542. [Google Scholar] [CrossRef]
  75. Siddiqi, D. (2018). Distributed Morphology. In J. Audring, & F. Masini (Eds.), The Oxford handbook of morphological theory (pp. 143–165). Oxford University Press. [Google Scholar]
  76. Smolensky, P., & Goldrick, M. (2016). Gradient symbolic representations in grammar: The case of French liaison [Unpublished manuscript]. Microsoft Research AI and Johns Hopkins University & Northwestern University. [Google Scholar]
  77. Starke, M. (2009). Nanosyntax: A short primer to a new approach to language. Nordlyd, 36, 1–6. [Google Scholar]
  78. Svenonius, P. (2012). Spanning [Unpublished manuscript]. CASTL, University of Tromsø. [Google Scholar]
  79. Taraldsen Medová, L., & Wiland, B. (2019). Semelfactives are bigger than degree achievements: The nanosyntax of Czech and Polish semelfactive and degree achievement verb stems. Natural Language & Linguistic Theory, 37, 1463–1513. [Google Scholar]
  80. Thomadaki, E. (1994). Μορφολογικά προβλήματα της Νεοελληνικής: H κλίση του ουσιαστικού [Morphological problems in Modern Greek: Noun inflection] [Ph.D. Thesis, National and Kapodistrian University of Athens]. [Google Scholar]
  81. Triantafyllidis, M. (2018). Νεοελληνική γραμματική (της Δημοτικής) [Modern Greek grammar (of the demotic variety)]. Institute of Modern Greek Studies. (Original work published 1941). [Google Scholar]
  82. Warburton, I. P. (1973). Modern Greek verb conjugation: Inflectional morphology in a transformational grammar. Lingua, 32, 193–226. [Google Scholar] [CrossRef]
  83. Zimmermann, E. (2018). Symbolic Representations in the Output: A case study from Moses Columbian Salishan stress. In S. Hucklebridge, & M. Nelson (Eds.), Proceedings of the forty-eighth annual meeting of the North East Linguistic Society (NELS 48) (pp. 275–284). GLSA. [Google Scholar]
  84. Zimmermann, E. (2021). Faded copies: Reduplication as distribution of activity. Glossa: A Journal of General Linguistics, 6(1), 58. [Google Scholar]
Table 1. Distribution and allomorphy pattern of stem-final /o/, /a/, and /i/.
Table 1. Distribution and allomorphy pattern of stem-final /o/, /a/, and /i/.
Stem-Final SegmentDistributionAllomorphy Pattern
Grammatical
Gender
ReferentStress
/o/FEM/MASC/NEUT±humanU/PU/APU-o/-Ø
/a/FEM/MASC±humanU/PU/APU-a/-Ø
/i/FEM/MASC±humanU/PU/APU-i/-Ø
Table 2. Distribution and allomorphy pattern of stem-final /a(ð)/, /i(ð)/, and /a(t)/.
Table 2. Distribution and allomorphy pattern of stem-final /a(ð)/, /i(ð)/, and /a(t)/.
Stem-Final Segment(s)DistributionAllomorphy Pattern
Grammatical
Gender
ReferentStress
/a(ð)/MASC (mostly)+human (mostly)U (SG), PU (PL)-á/-áð
/i(ð)/MASC+humanPU (SG), APU (PL)-i/-
/a(t)/NEUT–humanAPU-a/-at
Table 3. Comparison between two groups of stem endings.
Table 3. Comparison between two groups of stem endings.
Structure/Feature/o/, /a/, /i//a(ð)/, /i(ð)/, /a(t)/
Grammatical genderany valuefixed value
Referent (±human)any valuefixed value
Stressany positionfixed position
Allomorphy patternV-ØV-VC
Table 4. Distribution of stress patterns in three-syllable masculine nouns with the Theme exponents /o/ and /i/ based on Apostolouda (2018).
Table 4. Distribution of stress patterns in three-syllable masculine nouns with the Theme exponents /o/ and /i/ based on Apostolouda (2018).
Theme ExponentDatabaseAPUPUU
/o/ALex160
55.74%
38
13.24%
89
31.01%
TC79
43.16%
25
13.66%
79
43.16%
/i/ALex4
1.6%
180
72.28%
65
26.1%
TC1
0.69%
74
51.38%
69
47.91%
Table 5. Experimental results for stress assignment in three-syllable masculine nouns with the Theme exponents /o/ and /i/ based on Apostolouda (2018).
Table 5. Experimental results for stress assignment in three-syllable masculine nouns with the Theme exponents /o/ and /i/ based on Apostolouda (2018).
Theme ExponentAPUPUU
/o/144
45%
78
24.37%
98
30.62%
/i/69
21.56%
167
52.18%
84
26.25%
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Markopoulos, G. Breaking Down Greek Nominal Stems: Theme and Nominalizer Exponents. Languages 2025, 10, 85. https://doi.org/10.3390/languages10040085

AMA Style

Markopoulos G. Breaking Down Greek Nominal Stems: Theme and Nominalizer Exponents. Languages. 2025; 10(4):85. https://doi.org/10.3390/languages10040085

Chicago/Turabian Style

Markopoulos, Giorgos. 2025. "Breaking Down Greek Nominal Stems: Theme and Nominalizer Exponents" Languages 10, no. 4: 85. https://doi.org/10.3390/languages10040085

APA Style

Markopoulos, G. (2025). Breaking Down Greek Nominal Stems: Theme and Nominalizer Exponents. Languages, 10(4), 85. https://doi.org/10.3390/languages10040085

Article Metrics

Back to TopTop