Artificial word formation in the epic tradition: θοῦρος (‘fierce’) and the formula θούριδος ἀλκῆς

Abstract

The Homeric adjectives θοῦρος and θοῦρις (gen. -ιδος) are normally glossed as ‘rushing, impetuous, furious’. While θοῦρις exclusively qualifies feminine nouns, no feminine form of θοῦρος is attested. What was the exact relationship between θοῦρος and θοῦρις? In this paper it is argued that θοῦρις is not the paradigmatic feminine of θοῦρος, but an artificial formation of epic Greek. It arose in the formula θούριδος ἀλκῆς due to the metrical constraints of epic hexameter, and subsequently ousted the original feminine of θοῦρος. In elaborating this scenario, I show that the basic meaning of θοῦρος and θοῦρις is ‘fierce’. Other instances of artificial change of inflection in the Homeric Kunstsprache are discussed, and it is argued that the mechanism underlying their creation is linguistic contamination.

I. Introduction

The adjective θοῦρις is attested exclusively as an epithet of feminine nouns, and only occurs in hexameter verse and a few elegiac couplets. ¹ On the other hand, a feminine form of θοῦρος (whether in the shape θούρα/θούρη or θοῦρος) never occurs in Greek. In a way, then, θοῦρις functions as the feminine of θοῦρος. Why do we find only this abnormal feminine form? And how exactly do both stem forms relate to each other: could θοῦρις be an older paradigmatic feminine of θοῦρος, or is it a separate lexical item derived from θοῦρος?

When investigating these questions, it is imperative to first deal with the attestations and semantics of both forms, considering their contexts, the remarks of ancient scholia and lexicographers, as well as etymology (section II). Next, the Homeric evidence will be considered in more detail. I argue that θοῦρις spread from the formula θούριδος ἀλκῆς to its other occurrences, taking the place of the expected feminine, θούρη or θοῦρος (section III). Since verse-final *θούρης ἀλκῆς (or *θούρου ἀλκῆς) was unmetrical due to the metrical law called Meister’s Bridge, and since θοῦρις cannot be explained as an old paradigmatic feminine of θοῦρος (section IV), the question arises whether θούριδος ἀλκῆς could be an artificial creation. In section V, it is argued that θοῦρις is a metrically induced contamination between θοῦρος and its antonym ἄναλκις, and Homeric parallels for an artificial change of inflection are discussed. Finally, the view that θοῦρις is an artificial form presupposes that it cannot be explained as a ‘conventional’ morphological derivation. This claim is bolstered in section VI with an analysis of the functions of the suffixes -ι- and -ιδ- and their Indo-European precursors.

II. Attestations and lexical meanings of θοῦρος and θοῦρις

Already in the Greek lexicographical tradition, it was recognized that θοῦρις and θοῦρος belong together. ² Modern lexicographical works, in part following the ancient sources, generally consider θοῦρις and θοῦρος to be synonyms meaning ‘rushing, impetuous, furious’. ³ These translations are imprecise for several reasons. First, since epithets generally denote extra-temporal qualities of their referents rather than an ongoing action, a translation like ‘rushing’ does not make much sense. Moreover, renderings like ‘impetuous, i.e. that rushes/leaps with violent impetus at enemy’ ⁴ are not based on how θοῦρος is used, but inspired by the supposed etymological connection with θορϵῖν (‘to leap’) or by glosses given in the scholia. Finally, while the rendering ‘impetuous, furious’ is perhaps not impossible for θοῦρος, it is clearly inadequate for θοῦρις.

Let us consider the attestations of both forms in more detail. In Homer, θοῦρος is an exclusive epithet of the war god Ares (θοῦρον Ἄρηα 9x, verse final or before the trochaic caesura, and θοῦρος Ἄρης 2x). After Homer the adjective continues to be applied mainly to Ares (Tyrtaeus, Simonides, Euripides and also in an Attic epigram); ⁵ more exceptionally, it qualifies other warlike persons or beings, but only in the tragedians. ⁶ Instructive is the phrase καὶ θοῦρός πϵρ ὤν (Aesch. fr. 199.2 Radt, Prometheus speaking to Heracles). It is reminiscent of the Homeric (καὶ) κρατϵρός πϵρ ἐών (‘however fierce you are’) and shows that the meaning of θοῦρος was similar to that of κρατϵρός (‘fierce, warlike’). ⁷ Finally, θοῦρος does not qualify a person, but a weapon in θοῦρον … δόρυ (Eur. Rhes. 492, cf. also δόρυ θοῦρον in Ap. Rhod.).

The second form θοῦρις (gen. -ιδος) exclusively occurs in dactylic poetry and only modifies feminine nouns. It has an acc. sg. θοῦριν and is especially frequent in the verse-final phrase θούριδος ἀλκῆς (‘fierce fighting spirit’, 22x in Homer), which is preceded by various verb forms beginning at one of the third foot caesurae: μνήσασθϵ δὲ θούριδος ἀλκῆς (Il. 16.270, etc.), ἔπαυσϵ δὲ θούριδος ἀλκῆς (Il. 17.81) and several others. A number of these phrases also occur with χάρμης replacing θούριδος ἀλκῆς, for example, μνήσαντο δὲ χάρμης (Il. 4.222, etc.), παῦσϵ δὲ χάρμης (Il. 12.389). This gave epic poets the option of shortening (or extending) the phrase by one dactyl; to use the term coined by Egbert Bakker, χάρμης and θούριδος ἀλκῆς are ‘functional synonyms’ in such pairs of phrases. ⁸ This point illustrates that χάρμη and ἀλκή are near synonyms themselves: both express the concept of fighting spirit or martial courage. ⁹ In a second and more marginal use, θοῦρις in Homer qualifies the aegis and a shield, bringing to mind the post-Homeric phrase θοῦρον δόρυ just mentioned. ¹⁰ The derivative θούριος, which occurs mainly in the tragedians, likewise qualifies warlike fighters as well as their weapons; it is rightly considered to be a synonym of θοῦρος by the dictionaries. ¹¹

Now, an important question is: how could an epithet meaning something like ‘impetuous, furious’ be applied to a shield, an inanimate defensive weapon? The dictionaries and commentaries come up with various emergency solutions. ¹² Cunliffe remarks that there has been a ‘transference of epithet from the bearer to the shield’, but it is not immediately obvious what exactly is meant by this, other than that θοῦρος primarily qualified persons. ¹³ How could this transfer take place? It seems to me that sturdy weapons are called θοῦρος or θοῦρις because they are tough-looking: as such they confer ἀλκή on their bearer (who gains confidence by carrying them) and diminish that of an adversary (who will be intimidated by their aspect). ¹⁴ In this way we may also understand the use of ἄλκιμος (‘valiant, brave’) as a qualification of spears (ἄλκιμον ἔγχος, ἄλκιμα δοῦρϵ) in Homer.

In order to illustrate this, let us briefly review the three instances where the Homeric feminine acc. sg. θοῦριν qualifies a weapon. The first is ἀσπίδα θοῦριν (Il. 11.32), about Agamemnon picking up his shield in an arming scene, the place par excellence to describe weapons and their special properties. As becomes clear from its description, Agamemnon’s shield is intended to terrorize: ‘it was crowned with a wild-faced Gorgon, which gazed terribly and was flanked by Fear and Rout’ (Il. 11.36–37).

The phrase ἀσπίδα θοῦριν also occurs in a three-line description of Aeneas (Il. 20.161–63) trying to intimidate his opponent Achilles by showing his military prowess:

Αἰνϵίας δὲ πρῶτος ἀπϵιλήσας ἐβϵβήκϵι

νϵυστάζων κόρυθι βριαρῇ· ἀτὰρ ἀσπίδα θοῦριν

πρόσθϵν ἔχϵ στέρνοιο, τίνασσϵ δὲ χάλκϵον ἔγχος.

Aeneas had stepped out first, uttering threats, nodding with his heavy helmet; his fierce shield he held before his breast, and he brandished his bronze spear.

Finally, θοῦρις qualifies the aegis at Il. 15.308 (verse end αἰγίδα θοῦριν), an attribute of which the exact referent was probably no longer understood, but which is described in this passage as being carried ἐς φόβον ἀνδρῶν (‘for the routing of men’). ¹⁵

Thus, in all three passages θοῦριν qualifies a weapon or attribute that is explicitly described as having an intimidating effect. As a qualification of ἀλκή, θοῦρις may also have referred to the intimidating aspect of a warrior displaying his martial prowess. On the other hand, θοῦρος as an epithet of Ares signified not ‘furious’ (describing a state that applies within a specific time frame) but rather ‘fierce, warlike’ (a generic, timeless characteristic). I suspect that this is the older meaning and that the phrase θούριδος ἀλκῆς was interpreted by Homeric singers as describing the intimidating effect a fierce fighting spirit may have on opponents. In this sense, the epithet could then be applied to the shields of Agamemnon and Aeneas. ¹⁶

This is in part also what ancient commentators and lexicographers suggested. On the one hand, the scholia present us with clear folk etymologies, for instance when they gloss θοῦρον with θϵίως ὀρούων or θοῶς ὀρούων (‘who rushes like a god/quickly’) (for example, sch. vet. ad Il. 5.35, Erbse). This is a clear example of the strategy, known from Plato’s Cratylus, of etymologizing a word by providing it with a gloss that contains a permutation of all its constituent letters/sounds. On the other hand, we also find glosses that are concerned more with the sense, and they confirm what has just been said. The D-scholia ad Il. 5.30 render θοῦρον with ἐνθουσιαστικόν (‘raging, furious’) as well as ϵὐκίνητον ἐν τῷ πολέμῳ (‘agile in battle’). Various other sources gloss forms of θοῦρος or θοῦρις with πολϵμικός (‘warlike’) and/or ὁρμητικός, ¹⁷ which in this context is best interpreted as ‘warlike’, too. ¹⁸

Etymology, finally, is of little help in determining the lexical meaning of θοῦρος. A derivation from the root of θρῴσκω (aor. ἔθορον) (‘to jump, leap’) seems to be generally accepted. ¹⁹ This connection was already made by ancient lexicographers, commentators and in the scholia: for example, θούριδος ἀλκῆς· τῆς πολϵμικῆς. ἀπὸ τοῦ θορϵῖν, ὅ ἐστι πηδῆσαι, ἀφ’ οὗ καὶ θοῦρος Ἄρης (Hsch.). However, as we have seen, the translations ‘rushing’ or ‘jumping’ do not match the actual use of θοῦρος. In fact, the etymological relationship with θρῴσκω is by no means certain, as a viable alternative exists: θοῦρος could reflect an agentive nominal *d ^h ors-ó- (‘who ventures/attacks’), from the root of θρασύς (‘bold, dauntless’; from Proto-Indo-European (PIE)) *d ^h ers-, with loss of *-s- and compensatory lengthening. The formation would be of the same type as, for example, τομός (‘cutting, sharp’), φονός (‘who slays’). This etymology would account in a straightforward way for the lexical meaning of θοῦρος; it would presuppose that the accent of *d ^h orsó- was secondarily retracted. ²⁰

To conclude this lexicographical discussion, θοῦρος (‘fierce, warlike’) is mostly said of warriors, while θοῦρις means ‘fierce’ in the sense of ‘sturdy, tough-looking, intimidating’, qualifying weapons and fighting spirit. If there is a lexical difference between the two forms, it is slight: note that θοῦρος occurs as an epithet of δόρυ after Homer. Thus, lexical semantics gives us no compelling reason to expect a difference in stem formation.

III. The spread of θοῦρις in Homer

As a feminine form of θοῦρος, one expects either a form in -ᾱ (as normally in thematic adjectives) or in -ος (as usually in compounds, but also in some uncompounded thematic adjectives). ²¹ In reality, no feminine form of θοῦρος is attested in our extant evidence: it is θοῦρις which exclusively qualifies feminine nouns. Does it follow that θοῦρις is the paradigmatic feminine of θοῦρος? Not necessarily: θοῦρος is an uncommon poetic relic, and it typically qualifies referents of masculine gender. This means that the absence of a feminine form in, for example, Attic drama might in principle be ascribed to chance. ²² Moreover, occurrences of θοῦρις are limited to hexameter verse and elegiac couplets. Therefore, our question can be made more precise: why is the morphologically expected feminine form of θοῦρος (whether this was θούρη or θοῦρος) never attested in the epic tradition?

The key to the answer lies in the formulaic phrase θούριδος ἀλκῆς, which accounts for 22 of the 28 occurrences of θοῦρις in Homer. In verse-final position, *θούρης ἀλκῆς or *θούρου ἀλκῆς would not be well-formed because it violates Meister’s Bridge, which states that word end was avoided after a spondaic fifth foot. Of all the metrical laws concerning hexameter verse, Meister’s Bridge is observed best by far in Homer (better than Wernicke’s Law and Hermann’s Bridge). The exceptions, only a handful, are listed and discussed by Meister. ²³ A number of them are resolvable by the fact that contraction took place relatively late, notably in the formulaic phrase ἠῶ δῖαν for earlier *ἠόα δῖαν. ²⁴ Only three Homeric verse ends are irreducible counterexamples to Meister’s Bridge: κνῆ τυρόν (Il. 11.639), κρῖ λϵυκόν (Od. 4.604) and λὶς πέτρη (Od. 12.64). ²⁵ This means that the rule was very strict in earlier stages of the tradition.

I propose that the form θοῦρις first came into being as part of the phrase θούριδος ἀλκῆς, and that the other six Homeric instances, all in the accusative θοῦριν, were created as replacements of the older form (whether θοῦρον or θούρην) under the influence of this frequent formula. First, a few remarks on the acc. θοῦριν. Three of these six cases are found in the verse end θοῦριν ἐπιϵιμένοι ἀλκήν (‘clad in fierce fighting spirit’), said of the Aiantes. On two of these occasions the Aiantes are mentioned as part of a list of nine Achaean champions who volunteer to fight: τοῖσι δ’ ἐπ’ Αἴαντϵς θοῦριν ἐπιϵιμένοι ἀλκήν (Il. 7.164 and 8.262). ²⁶ This line may well be a very old one: ²⁷ the canon of Achaean champions (ἀριστῆϵς) is probably traditional in some form, and in both passages the well-known line mentioning Meriones, with its irregular scansion of Ἐνυαλίῳ ἀνδρϵιφόντῃ, follows two lines later. ²⁸

The formula θοῦριν ἐπιϵιμένοι ἀλκήν is remarkable for the brevis in longo in the final syllable of θοῦριν. This metrical irregularity is unexpected for two reasons. First, *θούρην ἐπιϵιμένοι ἀλκήν would be metrically regular and semantically unproblematic. Secondly, an alternative and metrically equivalent way of versifying a similar idea was available, as illustrated by the verse ἄνδρ’ ἐπϵλϵύσϵσθαι μϵγάλην ἐπιϵιμένον ἀλκήν (Od. 9.214, cf. also 514). ²⁹ Being metrically regular, this seems a more recent modification of the (traditional) verse mentioning the Aiantes.

It has been claimed that θοῦριν ἐπιϵιμένοι ἀλκήν is a case of metrical lengthening, ³⁰ but this does not take into consideration that this verse end may be a modification of an older, metrically regular prototype. ³¹ As is well known, metrical lengthening is virtually restricted to words of specific prosodic shapes, mainly tribrachic (for example, οὔνομα) and antispastic words (for example, Ἀπόλλωνα), and it could be applied secondarily in cretic sequences, too. ³² However, in words with a trochaic shape like θοῦριν, there was no motivation for metrical lengthening. One may therefore suspect that the prototype of this formula had the shape *θούρην ἐπιϵιμένον ἀλκήν. This may have been changed at some point into θοῦριν ἐπιϵιμένον ἀλκήν, in spite of the metrical irregularity that was introduced, because θοῦρις had come to be viewed as the standard epithet of ἀλκή due to the frequency of the phrase θούριδος ἀλκῆς.

Next, the same substitution could take place also in the verse-final phrases ἀσπίδα θοῦριν and αἰγίδα θοῦριν (which account for the remaining three occurrences), especially since this had no metrical repercussions. This second substitution was perhaps made easier by the fact that θοῦρος was otherwise used exclusively as an epithet of Ares with the meaning ‘fierce, warlike’, which made it less suited to defensive weapons (cf. section II). Since ἀλκή was conceptually close to a defensive weapon (it is something which protects, and in which one dresses), it was logical to generalize the stem form known from the formulaic phrases θούριδος ἀλκῆς and θοῦριν ἐπιϵιμένον ἀλκήν.

In these last paragraphs I have assumed that the original feminine form was θούρη, but as I remarked earlier, it cannot be excluded that θοῦρος once had this function, too. In fact, in that case it would be even easier to explain why, for example, ἀσπίδα θοῦρον was replaced by ἀσπίδα θοῦριν: the latter would have been more clearly marked as a feminine. On the other hand, if θοῦρος was the original feminine form, the metrical irregularity in θοῦριν ἐπιϵιμένοι ἀλκήν remains unaccounted for.

Thus, all Homeric occurrences of the accusative θοῦριν can be viewed as secondary replacements of either θούρην or θοῦρον under the influence of the frequent phrase θούριδος ἀλκῆς. We have seen that θούριδος was preferred in that phrase over θούρης or θούρου because these alternatives would have caused a violation of Meister’s Bridge. As I will argue in section V, poets probably also created the form θούριδος for this very reason. However, let us first see whether θοῦρις as a feminine of θοῦρος can be explained by normal morphological processes.

IV. θοῦρις is not a flexional feminine in origin

There are two conventional linguistic means to account for a feminine adjectival form θοῦρις, having approximately the same meaning as θοῦρος. It could be an old paradigmatic feminine of θοῦρος, or originally a separate adjective derived from θοῦρος. Most handbooks remain vague or agnostic about these possibilities. ³³ Is one of these scenarios likely?

The first option, an old paradigmatic feminine, was defended by Wackernagel. ³⁴ This may seem reasonable in view of the distribution between θοῦρος and θοῦρις, but there is one very serious objection. There are two regular ways in Greek to mark feminine agreement with adjectives (both directly reflecting a procedure inherited from Indo-European):

• with third declension stems: add the suffix *-i̯a, *-i̯ā-, for example, βαρύς, f. βαρϵῖα < *-eu̯-i̯a; φοῖνιξ, f. φοίνισσα < *-īk-i̯a;

• with stems in -ος: substitute *-ā- (PIE *-eh ₂ -) for the thematic vowel, for example, νέος, f. νέᾱ.

By contrast, the suffixes -ι- and -ιδ- are not normally used to mark the feminine of adjectives, but Wackernagel apparently thought that -ίδ- had this function in the prehistory of Greek. In support of this claim he cites the following Homeric words in -ίς:

• κορωνός adj. ‘curved’ (Hippoc.); ‘having a curved horn’ (Archil.) → κορωνίς f. (17x dat. pl. κορωνίσιν, formulaic epithet of νηυσί ‘ships’);

• ἥμϵρος adj. ‘tame, cultivated’ → ἡμϵρίς f. ‘cultivated vine’;

• νύκτϵρος adj. ‘of the night, nightly’ (Aesch.+) → νυκτϵρίς f., properly ‘(animal) of the night’, hence ‘bat’;

• κραταιός adj. ‘strong, powerful’ → κραταιίς f. ‘powerful force’ (Wackernagel speaks of a ‘substantiviertes Femininum’). ³⁵

Before continuing, we must briefly address the accentuation and inflection of stems in -ις. If correctly adduced, the above parallels (which have the oxytone suffix -ίς, -ίδος irrespective of the accent of the base form) would lead one to expect a form *θουρίς, but instead we find θοῦρις, with barytone accentuation and an accusative θοῦριν. Wackernagel accounted for this as a case of ‘Aeolic barytonesis’: traditional epic words with an unexpectedly retracted accent that stem from an Aeolic dialect with recessive accentuation. In this context, he mentioned other forms such as ἵππουρις and ἄκοιτις; he saw the acc. sg. in -ιν of these words as typical for Aeolic. ³⁶ However, the evidence for Aeolic barytonesis in Homer is not abundant, and both the barytone accentuation of these compounds and their acc. in -ιν may be seen as archaisms. ³⁷ On the other hand, the accentuation of θοῦρις may have been influenced by that of θοῦρος, and it sometimes happens in Homer that an acc. sg. in -ιν is secondary for -ιδα. ³⁸ This means that θοῦρις can, but need not, be compared to forms in -ίς, -ίδος (that is, the accentuation and inflection of θοῦρις do not directly inform us about the type of derivation).

Returning to the above examples: do they justify Wackernagel’s suggestion that -ίδ- could originally mark the feminine of adjectives in -ος? Upon closer consideration, none of the cases is convincing. ³⁹ There is no reason to think that the nouns ἡμϵρίς, νυκτϵρίς and κραταιίς were once part of the paradigms of the adjectives ἥμϵρος, νύκτϵρος and κραταιός. Moreover, κραταιός has its own feminine, attested in the old formula μοῖρα κραταιή. If anything, ἡμϵρίς, νυκτϵρίς and κραταιίς are substantivizations derived from their respective adjectival base forms. ⁴⁰ By comparison, the most promising parallel for θοῦρις as a feminine of θοῦρος is the epithet κορωνίς, but the derivational analysis of this form remains uncertain, as we will see in section VI.

Thus, adjectives in -ος never had a paradigmatic feminine in -ίς. An alternative account would be that θοῦρις was originally an independent derivative that was secondarily incorporated into the paradigm of θοῦρος as its feminine form. As we will see in section VI, such an account cannot be easily applied to the pair of forms θοῦρος with θοῦρις either. Before going into the technical issues, I will first show how θοῦρις may have originated as an artificial substitute for the feminine of θοῦρος.

V. Θούριδος ἀλκῆς as a case of artificial contamination

As we have seen, the formula θούριδος ἀλκῆς stands in for metrically awkward *θούρης ἀλκῆς or *θούρου ἀλκῆς. I submit that the ending -ιδος in this formula arose under the influence of ἄναλκις -ιδος ‘cowardly; coward’ (20x Hom.), an antonym of θοῦρος. Later on, the new feminine θοῦρις, -ιδος completely ousted the original form according to the scenario proposed in section III.

Before discussing how ἄναλκις -ιδος influenced the creation of θούριδος ἀλκῆς, let us first consider some well-known Homeric parallels for a change of declension or conjugation type, ⁴¹ so as to better understand the nature of artificial word formation in Homer.

Beside the formulaic phrase τϵτιηότι θυμῷ (‘with a vexed spirit’, instrumental dative), we find another formulaic phrase τϵτιημένος ἦτορ (‘vexed at heart’, where ἦτορ is an internal accusative). The reason for changing the active participle into a middle one is that *τϵτιηὼς ἦτορ would have violated Meister’s Bridge. ⁴² Replacement with a middle form was made easier by the significant semantic overlap between both voices in the perfect stem, and also by the fact that verses ending in -μένος ἦτορ were more widespread (for example, βϵβολημένος ἦτορ, κϵχολωμένος ἦτορ).

The middle voice could be used artificially also in the present or aorist stem in cases where using the active voice would have violated Meister’s Bridge or Wernicke’s Law. A clear example is furnished by the phrases καὶ ἐξϵρϵϵίνϵτο μυθῷ (Il. 10.81) and ἀφὰρ δ’ ἐρϵϵίνϵτο μυθῷ (Od. 17.305), both in speech introductions, replacing *ἐρέϵινϵν μυθῷ. ⁴³ The sequence -ϵτο μύθῳ occurring in other verse ends may have served as a model: cf. καὶ ἐπϵίθϵτο μύθῳ (Il. 1.33, 24.571) and especially καὶ ἀμϵίβϵτο μύθῳ (Il. 24.200 and passim), again in a speech introduction.

The name Ἀντιφάτης occurs five times in Homer, with an accusative Ἀντιφάτην (twice) and a verse-final genitive Ἀντιφάταο (Od. 10.106). The accusative also appears in the form Ἀντιφατῆα (Od. 10.114), belonging to a would-be nominative Ἀντιφατϵύς but referring to the same Ἀντιφάτης as Ἀντιφάταο eight lines before. We must assume that the poet used the accusative of a name in -ϵύς for metrical reasons. From a morphological viewpoint this was certainly not trivial, because nouns in -ϵύς are normally not compounds, as noted by Meister. ⁴⁴ Similar instances are found in the inflection of compounds in -ος. For instance, the genitive of ἡνίοχος appears in the form ἡνιόχοιο, used five times at verse end. The wish to use other case forms in the same metrical position then led poets to create the artificial forms ἡνιοχῆα and ἡνιοχῆϵς. ⁴⁵

Perhaps the most famous cases of artificial nominal inflection in the fifth foot are the verse-final phrases ϵὐρέα πόντον and ϵὐρέα κόλπον. They are artificial because the only current form of the masculine acc. sg. ending of adjectives in -ύς in any form of Greek is -ύν, not -έα. The motive for avoiding the grammatically regular phrases ϵὐρὺν πόντον and ϵὐρὺν κόλπον is that these would have violated Meister’s Bridge. ⁴⁶ Witte proposed that ϵὐρέα πόντον (2x Il., 1x Od.) arose by inflection of the dative ϵὐρέϊ πόντῷ (7x Od.), that is, by analogical levelling of the paradigm, and in this he is all but universally followed. ⁴⁷ However, he does not explain how poets were able to create an ungrammatical acc. sg. ϵὐρέα. In my view, it is more attractive to suppose that the ending of ϵὐρύν was artificially changed into that of an adjective in -ής, a type which shares many case forms with stems in -ύς. This is made likely by the existence of other verse ends in -έα πόντον, notably ἐπ’ ἠϵροϵιδέα πόντον. ⁴⁸

What actually happened in linguistic terms in these examples is usually left unstated. ⁴⁹ In my view, all of the artificial epic forms discussed above are best viewed as contaminations or blends with other forms and formulas that came to the poets’ minds as they composed. ⁵⁰ As is stressed by treatments of contamination in handbooks about historical linguistics, ⁵¹ contamination usually occurs in words that regularly co-occur, for example, in antonymic pairs (cf. English male and female for older male and femelle) or in sequences of numerals (for example, in the Greek dialect of Heraclea, where ὀκτώ, ἐννέα become hοκτώ, hϵννέα under the influence of ἕξ, ἑπτά). In epic verse composition, contaminations are likely to have been promoted by the fact that specific words regularly occurred in certain metrical positions. The examples just discussed clearly illustrate this point. Thus, in the case of verse-final ἡνιοχῆα, -ῆϵς, -ῆας, some poet may have created these forms when he was reminded of the simplex noun ὀχϵύς (13x Hom.), which occurs in the case forms ὀχῆα, ὀχῆϵς, ὀχῆας (12x in total), and always in verse-final position. A similar example is πατροφονῆα (3x Od., each time referring to Aegisthus): we may suspect that the poet of the Odyssey (or a precursor) coined it by crossing the simplex φονϵύς (‘slayer’, Hom.+, attested at verse end in gen. sg. φονῆος, acc. pl. φονῆας) with the expected but rare compound πατροφόνος (‘slaying a father’), or with the more usual πατροκτόνος (‘parricide’).

In epic Greek, conditions were favourable for a contamination between θοῦρος and ἄναλκις. ⁵² The demands of verse composition provided a clear motivation for reshaping existing word forms. When confronted with the undesirable metrical shape of verse-final *θούρης ἀλκῆς or *θούρου ἀλκῆς, epic poets may have been led to think of phrases containing the antonym ἄναλκις. ⁵³ This connection was favoured in particular by the fact that ἄναλκις is linked to ἀλκή etymologically. In addition, Homer uses the phrases ἀνάλκιδα θυμόν (‘cowardly spirit’, Il. 16.355 and 656) and ἀνάλκιδα φύζαν (‘weak-spirited panic’, Il. 15.62), which refer to the opposite of θούριδος ἀλκῆς (‘fierce fighting spirit’). These are similar not only in terms of semantics, but also word order and word shape: in all cases the head noun follows the modifier, and it is disyllabic. ⁵⁴ Finally, as one reviewer of this paper remarked, the fact that θ…ιδ- ἀλκ- (in θούριδος ἀλκῆς) is a phonic permutation of -άλκ-ιδ- θ… (in the phrase ἀνάλκιδα θυμόν) may have played a role.

The proposed source of the contamination, ἄναλκις -ιδος, is an adjective in -ι(δ)- with identical masculine and feminine forms. Such adjectives were on the verge of extinction (cf. section VI.ii), and the suffix -ιδ- in the contaminated phrase θούριδος ἀλκῆς could easily be reinterpreted as a feminine marker: nouns with this suffix were feminine almost without exception, and there were a few other adjectives in -ος with a derived feminine in -ίδ- (cf. sections III and VI.i). Consequently, other forms of the original feminine of θοῦρος could also be replaced by forms of θοῦρις along the lines set out in section II.

Two further epic words in -ίς must be discussed as parallels for an artificial contamination: the nouns παλλακίς (‘concubine’) and ἀγκαλίς (‘bent arms, embrace’). As noted by Meier, ⁵⁵ the use of both forms in Homer must be metrically conditioned: compared to the Classical Ionic-Attic forms παλλακή and ἀγκάλη, they appear to have undergone an artificial change of declension type: ⁵⁶

• Since the only practicable case forms in a hexameter were παλλακή, παλλακῇ and παλλακαί (before vowels), whereas παλλακίς could be used in all case forms, it is likely that the latter is an artificial substitute for παλλακή. ⁵⁷ In my view, it is conceivable that παλλακή was blended with concubine names in -ίς such as Χρυσηΐς, Βρισηΐς, perhaps aided by ἄκοιτις (‘spouse’) (though that has a different accentuation).

• In Homer, ἀγκαλίς occurs only in the phrase ἐν ἀγκαλίδϵσσι (Il. 18.556, 22.503), which clearly stands in for ἐν ἀγκάλαις (unmetrical in a hexameter, but common in Classical Attic). Outside of hexametrical Greek, ἀγκαλίς does occur, but with a different meaning, ‘armful’, that is, ‘that which is carried in bent arms’ (for example, Ar. fr. 418). In fact, in Il. 18.556 both meanings are conceivable, and this ambiguity may have licensed the use of ἐν ἀγκαλίδϵσσι as a substitute for ἐν ἀγκάλαις (‘in the arms’) at Il. 22.503.

The examples discussed in this section illustrate that the idea of an inflectional contamination of θοῦρος with ἄναλκις within epic Greek is an attractive option, for which various parallels can be adduced.

VI. Accounting for the pair of forms θοῦρις with θοῦρος by means of derivation

In this final section, I argue that θοῦρις cannot have been derived from θοῦρος in a regular way, neither by derivational mechanisms that were operative within Greek, nor as an archaism inherited from the Indo-European parent language.

i. θοῦρις derived from θοῦρος within Greek

Most handbooks assume a derivational relation between θοῦρις and θοῦρος but are vague about the precise details. An exception is the overview of Greek derivational morphology by Balles, ⁵⁸ who categorizes θοῦρος → θοῦρις as an adjectival feminine ‘Motionsbildung’ (i.e. θοῦρις would have been introduced to distinguish the natural sex of the referent). This is not plausible, as θοῦρις does not qualify natural female beings.

In order to appreciate the problem more clearly, let us briefly look at the different types of feminine forms with the suffix -ίδ-. ⁵⁹ The main issue is that most words in -ίδ- are nouns, whereas θοῦρις is an adjective. There are, however, instances where -ίδ- does make adjectives. Consider the following types of derivation: ⁶⁰

1. 1. feminine patronymics: ὁ Ἄτλας → ἡ Ἀτλαντίς (‘daughter of Atlas’);

2. 2. feminines of ethnic names: ὁ Δάρδανος → ἡ Δαρδανίς (‘Dardanian woman’);

3. 3. feminine nouns identifying the referent by a toponym: οἱ Δϵλφοί → ἡ Δϵλφίς (‘woman from Delphi, the territory of Delphi, a coin from Delphi’);

4. 4. feminine adjectives denoting appurtenance: ὁ βασιλϵύς → f. adj. βασιληΐς (‘kingly, royal’) (Hom.+); ὁ στρατιώτης → f. adj. στρατιῶτις (‘of a soldier’) (Classical); ⁶¹

5. 5. feminine nouns derived from masculine nouns in -της: ὁ πολίτης → ἡ πολῖτις (‘female citizen’); ⁶²

6. 6. feminine nouns for objects characterized by the base word: τὸ ἄργυρον → ἡ ἀργυρίς (‘silver cup’);

7. 7. diminutives: ἡ ἅμαξα → ἡ ἁμαξίς (‘little cart’).

It is debatable how all these different derivational types are interrelated. Some of them (types 5 and 7) are clearly secondary. The function of deriving lexicalized nouns for concrete objects (type 6) is well-established, but irrelevant for θοῦρις and perhaps unrelated to types 1 to 4. ⁶³

More relevant for present purposes are the adjectives of appurtenance (type 4). It is likely that types 1, 2 and 3 were originally adjectives of appurtenance as well. ⁶⁴ An illustrative example is Αἰολίς (‘Aeolian’), which functions as a patronymic denoting a daughter of Aeolus, as a feminine adjective of appurtenance (used of cities, dialect and the musical mode), and as a noun denoting the region Aeolid. It is possible that the feminine of ethnic names (type 2) developed from the patronymic use: Δαρδανίς may have originally referred to a female descendant of the eponymous ancestor Δάρδανος.

It appears, however, that θοῦρις cannot be analysed as an original adjective of appurtenance. First of all, the supposed base form θοῦρος is not a noun (as in all the above cases) but an adjective. ⁶⁵ In response to this, one might hypothetically assume that θοῦρις was derived from a now-lost noun, for example, *θουρά or *θοῦρος (‘battle rage’). However, a second problem is that adjectives of appurtenance in -ίς are derived from nouns belonging to specific semantic categories: personal names, place names and words denoting concrete persons or objects. ⁶⁶ This also holds for the following two feminine adjectives in -ίς, attested in Classical historiography, which offer a further illustration of the problem:

• συμμαχίς (‘allied, ally’) (Thuc., Hdt.) beside σύμμαχος -ον (‘allied, ally’)

• πϵριοικίς (‘neighbouring, neighbour’) (Thuc., Hdt., Xen.) beside πϵρίοικος (‘neighbouring, neighbour’)

These cases have in common with the pair of forms θοῦρις with θοῦρος that the special feminine form is derived from an adjective in -ος. That is where the similarity ends, however. Deriving the forms συμμαχίς and πϵριοικίς was possible because σύμμαχος and πϵρίοικος usually refer to persons and may be used as substantives. Indeed, συμμαχίς and πϵριοικίς are often used in apposition to nouns with concrete referents (πόλις, ναῦς and πόλις, νῆσος, χώρα, respectively), and this is precisely why ἡ συμμαχίς may occurs in substantivized form, denoting an allied military force. ⁶⁷ The semantics of θοῦρις ἀλκή is quite different: it does not have a concrete referent but denotes an abstract entity. Therefore, a hypothetical substantivization ἡ θοῦρις would hardly be meaningful, as it would not distinguish a concrete individual specimen of fighting spirit.

We may conclude that θοῦρις cannot be derived from θοῦρος as a feminine adjective.

ii. The pair of forms θοῦρις with θοῦρος as an Indo-European inheritance

Another function of -ίδ- was to transform an adjective in -ος into a feminine noun, as in νύκτϵρος (‘nightly’) → νυκτϵρίς (‘bat’) (cf. section III). This pattern of derivation differs from that discussed in section VI.i in that it operates on adjectives. In a recent proposal, Nussbaum has argued that θοῦρος → θοῦρις is an instance of this substantivizing derivation. ⁶⁸

In this function, -ίδ- was probably an extended form of the suffix -ι- inherited from Proto-Indo-European. The exact origin and function of these adjectival stems in -i- is a subject of debate, ⁶⁹ but a widely followed scenario was formulated by Schindler and further elaborated by Nussbaum. These scholars have argued that the suffix *-i-, when replacing *-o- in adjectives, made adjectival abstracts. ⁷⁰ More recently, Nussbaum has argued that *-i- could also form substantivizations of adjectives in Proto-Indo-European. ⁷¹ The evidence includes derivations like the following:

• adjective in * -o- → substantivization in -i-

• Lat. adj. rauus (‘hoarse’) → rauis (‘hoarseness, angina’)

• Gr. adj. ἄκρος (‘high, top’) → ἄκρις (‘summit’) ⁷²

• Vedic Sanskrit adj. jīrá- (‘agile, lively’) → jīrí- (‘flowing water’)

• Avestan adj. tiγra- (‘sharp, pointed’) → tiγri- (‘arrow’)

If this scenario is correct, how should we account for the existence of adjectives in -i- in Greek and other Indo-European languages, especially in Latin where adjectives in -is are widespread (cf. fortis, grauis, etc.)? ⁷³ Nussbaum views these adjectives in -i- as ‘re-adjectivizations’ of substantivized nouns. ⁷⁴ Applying this scenario to θοῦρις, he analyses this as a substantivized form in origin, meaning ‘(the) fierce one’. At some point, θοῦρις would have been pushed back into the function of a regular adjective, and then reinterpreted as the feminine of θοῦρος. Nussbaum compares θοῦρις with the Homeric epithet of ships, κορωνίς, and speaks of both forms as ‘inconsistent and abortive creations of specifically feminine adjectival forms’. ⁷⁵

Does this scenario account for the actual use of these feminines in -ις? As for κορωνίς, this epithet may mean something like ‘curved’ or ‘having a curved bow’, ⁷⁶ but its derivation from κορωνός remains uncertain. After Homer the form is used not as an adjective, but as a noun denoting concrete objects that are characterized by a curved shape or curved parts. ⁷⁷ We must also take into account that the related noun κορώνη (a substantivization of κορωνός) may denote the curved bow of a ship. ⁷⁸ Therefore, it is conceivable that κορωνίς, in the Homeric phrase νηυσὶ κορωνίσιν, denotes a type of ship characterized by its κορώνη. ⁷⁹ If κορωνίς was derived not from κορωνός but from κορώνη, it is similar to other de-substantival ship names in -ίς attested in the Classical period; ⁸⁰ and comparable to the Homeric φόρτος (‘cargo’) → φορτίς (‘freighter’).

There are also a handful of other Greek adjectives in -ις which Nussbaum does not comment on, but which may well be analysed as i-stem substantivizations of o-stem adjectives. ⁸¹ The most important instances attested before the Classical period are: ⁸²

1. 1. ἴδρις, usually translated as ‘skilled, skilful, experienced’ (Hom.+). In Homer and Hesiod, the syntactic behaviour of ἴδρις is consistent with that of a noun or substantivized adjective meaning ‘expert, skilled person’. ⁸³ In the following lines (Od. 6.232–34 = 23.159–61), ἴδρις is used in apposition to ἀνήρ, as a runover word in enjambment:

   ἀνὴρ

   ἴδρις, ὃν Ἥφαιστος δέδαϵν καὶ Παλλὰς Ἀθήνη

   τέχνην παντοίην.

   a man, an expert, whom Hephaestus and Pallas Athena have taught a versatile craft.

   The only other Homeric occurrence is Φαίηκϵς … ἴδριϵς … νῆα θοὴν ἐνὶ πόντῳ ἐλαυνέμϵν (‘the Phaeacians … experts in steering a swift ship on sea’, Od. 7.108–09). Meusel has already shown that the adjectival use of ἴδρις may be a late development, starting from its use in apposition to ἀνήρ or γυνή. Moreover, he argues that Pindar may preserve a trace of an older abstract noun ἴδρις (‘expertise’). ⁸⁴

2. 2. ϵὖνις (‘bereft, severed’) (Hom.+). As I argue elsewhere, ⁸⁵ ϵὖνις is originally a substantivization meaning ‘bereft person’, as in both its Homeric attestations it is used predicatively as the object of a so-called light verb meaning ‘to make, render’: ὅς μ’ υἱῶν πολλῶν τϵ καὶ ἐσθλῶν ϵὖνιν ἔθηκϵ (Il. 22.44) and ψυχῆς τϵ καὶ αἰῶνός … ϵὖνιν ποιήσας (Od. 9.523–24).

3. 3. ψϵῦδις (‘lying, deceitful’) (hapax, Pind. Nem. 7.50). In its sole attestation, ψϵῦδις is used predicatively and might in fact be analysed as a noun: οὐ ψϵῦδις ὁ μάρτυς ἔργμασιν ἐπιστατϵῖ (‘it is no lying witness that presides over achievements of the offspring of you and Zeus, Aegina’, tr. Race (1997)).

4. 4. τρόφις (‘full-grown’) vel sim. (Il. 11.307; Hdt. 4.9.4), a complicated case. ⁸⁶ In its sole Homeric attestation, which is of disputed interpretation, it might perhaps be a neuter noun: πολλὸν δὲ τρόφι κῦμα κυλίνδϵται (Il. 11.307). ⁸⁷ The predicative use in ἐπϵὰν γένωνται τρόφιϵς (Hdt. 4.9.4) is compatible with an old noun meaning something like ‘grown-up, adult’.

In cases like ἴδρις and ϵὖνις we might indeed speak of the ‘abortive creation’ (to borrow Nussbaum’s term) of an adjective from its substantivized form. ⁸⁸ We must now ask whether θοῦρις can be analysed in the same way. The answer is negative, for two reasons. First of all, whereas the other four simple adjectives in -ις discussed above are used predicatively or in apposition in their oldest attestations, θοῦρις is used attributively in all its Homeric attestations. Secondly, the other forms in -ις have concrete referents (in most cases, persons) of which they describe an acquired and distinctive characteristic, while the epithet θοῦρις modifies an abstract noun (ἀλκή, ‘fighting spirit’) of which it describes an inherent, generic aspect. Thus, the assumption that θοῦρις is an old substantivization appears to be gratuitous.

Let us finally discuss the origin of ἄναλκις. The synchronically expected derivation from ἀλκή would be *ἄναλκος. It is widely assumed that the suffixation of ἄναλκις is due to a prehistoric rule stating that *-o- had to be replaced by *-i- in possessive compounds. This compound substitution is indeed widespread in Latin (e.g. arma → inermis, lingua → trilinguis, etc.) and in certain Celtic languages, but traces of the same rule in Indo-Iranian and Greek are meagre at best. ⁸⁹ In fact, ἄναλκις is the only Greek example usually mentioned in this connection. Moreover, in Latin, simplex i-stem adjectives were also widespread (for example, fortis, grauis, etc.). In my view, it is more attractive to account for ἄναλκις and a few other Greek compounds by the same token as simple adjectives in -ις: they are originally substantivizations of adjectives in -ος. In other words, ἄναλκις was originally a noun meaning ‘one without ἀλκή, coward’. It would have competed with an adjective *ἄναλκος ‘without fighting spirit’ that was used attributively. When the distinction between -ος and -ις became obsolete, most forms in -ις lost currency, but ἄναλκις (‘coward’) was preserved, presumably because it was much more frequent than its attributive counterpart *ἄναλκος. ⁹⁰ Indeed, in Homer ἄναλκις is often used predicatively as an invective (Il. 9.34–35): ⁹¹

ἀλκὴν μέν μοι πρῶτον ὀνϵίδισας ἐν Δαναοῖσι

φὰς ἔμϵν ἀπτόλϵμον καὶ ἀνάλκιδα.

[F]or first of all you have taunted my fighting spirit among the Achaeans, saying that I am unwarlike and a coward.

On the other hand, in three Homeric instances ἄναλκις does function as an attributive adjective meaning ‘cowardly’: ἀνάλκιδα θυμόν (Il. 16.355 and 656) and ἀνάλκιδα φύζαν (Il. 15.62). Thus, ἄναλκις covers functions that (we may suppose) were once fulfilled by *ἄναλκος. ⁹²

Two other old compounds in -ις can be analysed along the same lines. First, ἄκοιτις ‘spouse’ looks like the substantivization of an older adjective *ἄκοιτος < *sm̥-koito- (‘who shares the same bed’), derived from κοῖτος (‘sleep, bed’) or κοίτη (‘bed’). ⁹³ Secondly, Homeric ἵππουρις is normally considered an adjective meaning ‘with a horse-tail crest’ (cf. οὐρά, ‘tail’), qualifying words for ‘helmet’. However, in six of its seven Homeric occurrences, ἵππουρις is a runover word starting the line, and clearly placed in apposition. As with κορωνίς, we may suspect that it was not merely an epithet but denoted a type of helmet, the ἵππουρις. Thus, ἵππουρις would be the substantivization of an adjective *ἵππουρος (‘having a horse tail’).

In sum, the other adjectival stems in -ις are mostly used predicatively or placed in apposition, and can therefore be analysed as old substantivizations. ⁹⁴ The form κορωνίς is used attributively in the phrase νηυσὶ κορωνίσιν, but it modifies a noun with a concrete referent. By contrast, the syntactic use and lexical meaning of θοῦρις are very different: there is no indication that it originated as a predicative or substantival form, and it primarily modifies an abstract noun, ἀλκή. Indeed, the only other adjective in -ις that modifies abstract nouns is ἄναλκις (θυμός, φύζα) – another point in favour of the contamination proposed in section V.

VII. Conclusions

The Homeric adjective θοῦρις -ιδος is best rendered as ‘fierce’ (of fighting spirit) and ‘tough, intimidating’ (of weapons). It functions as the feminine of θοῦρος, which also means ‘fierce’ and is an epithet of Ares. Since -ις was not originally a morphological marker of the feminine of adjectives, the pairing of θοῦρις with θοῦρος must be accounted for.

A derivational suffix -ις does occur in various different functions: a few other Greek adjectives in -ος have a special feminine form in -ίς, and there are parallels from other Indo-European languages for a suffix replacement -o- → -i- making substantivized adjectives. However, these cases are all quite different from the case of θοῦρος with θοῦρις. First, other adjectives in -ις are mostly used predicatively or in apposition, while θοῦρις is only used attributively. Secondly, other instances of -ίς beside -ος have concrete referents, whereas θοῦρις qualifies the abstract noun ἀλκή.

I have therefore proposed a scenario in which θοῦρις was created as an artificial form, through contamination of θοῦρος with the antonym ἄναλκις -ιδος (‘cowardly’), whose i-stem inflection is an archaism. This contamination first occurred in the traditional verse-final formula θούριδος ἀλκῆς, in avoidance of unmetrical *θούρης ἀλκῆς or *θούρου ἀλκῆς (Meister’s Bridge). Subsequently, the accusative θοῦριν supplanted the original form (*θούρην or *θοῦρον) in other phrases. If the original form was *θούρην, this may explain the occurrence of brevis in longo in the traditional phrase θοῦριν ἐπιϵιμένοι ἀλκήν.

This study has shown how attention to morphological detail may shed light on issues of lexical semantics, on the syntactic behaviour of adjectives and on the mechanisms governing the creation of formulaic phrases in the epic tradition. It has been argued in passing that many well-known artificial forms in Homer are best analysed as contaminations. If the scenario proposed here for θούριδος ἀλκῆς is correct, it also offers further support for the antiquity of Meister’s Bridge, and for the creation of artificial word forms at the early stage of the epic tradition when the phrases containing θούριδος ἀλκῆς were coined.