BMCR 2007.02.05

s-stem nouns and adjectives in Greek and Proto-Indo-European. A diachronic study in word formation

, s-stem nouns and adjectives in Greek and Proto-Indo-European. A diachronic study in word formation. Oxford: Oxford University Press, 2006. xii, 264. £50.00.

1 Responses

Table of Contents

This book is a good example of sound philological scholarship that prevents us from reconstructing ad infinitum. Dealing with the s-stems within Greek and from an Indo-European perspective, it offers many useful insights both to Indo-Europeanists and to scholars of Greek. The former are made aware that not every queer-looking word in Greek must be inherited from the proto-language while the latter are reminded that some riddles of individual languages still are only to be solved by comparison with their sister languages.

The book is divided into four chapters preceded by a prologue and followed by a short epilogue. The first chapter (pp. 6-44) gives an exhaustive history of the research done in the area, especially in connection with the so-called ‘Caland system’ of suffixes; the second chapter deals with the neuter s-stem nouns of the type γένος (pp. 45-128); the third chapter with animate s-stems of the type αἰδώς (pp.129-159); and the fourth chapter treats s-stem adjectives of the type εὐμενής and ψευδής (pp. 160-215). The concluding index makes all principal roots and words discussed easily accessible (pp. 243-264). There are very few misprints and I came across only one major blunder on p. 155 where “Hitt. * w < *- r” should probably read “Hitt. watar < *-r".

In the first chapter M. gives a short overview of pre-neo-grammarian treatments of s-stems in works of scholars like Bopp and Schleicher, and most importantly calls attention to Léon Parmentier, who in his 1889 work already made many observations that M. himself subscribes to in the following chapters. Contrary to the common theory of the time, Parmentier observed that not all s-stem adjectives seem to be built on s-stem nouns as in the case of μένος : εὐμενής. E.g., δισθανής‘twice dead’ seems to correspond to the aorist ἔθανον, whereas a noun ** θάνος does not exist. Under this assumption one is freed from the inconvenience of assuming that the supposedly underlying nouns on which the adjectives were built have all disappeared for no good reason. Secondly, Parmentier followed Saussure in the rejection of Brugmann’s hypothesis that the Greek pair ψεῦδος : ψευδής reflects an IE type reflected in Sanskrit in the type távas-‘strength’ : tavás-‘strong’, as well. Against this he argued that there are no word equations between IE languages for simple s-stem adjectives; that, if the oxytonesis as seen in Greek ψευδής were old, one would expect the zero-grade in the root (i.e. ** ψυδής); and that simple s-stem adjectives are rare in Greek and Sanskrit (no ** ἡδής, κρατής, θαρσής, etc., instead we find u-stem adjectives). Finally, Parmentier already noted the relationship between s-stem neuters and u-stem and ro-stem adjectives as in κράτος, κρατύς and κῦδος, κυδρός. Thus, apart from the observation made by Caland that in compounds an i-stem is used as first member instead of an -s- or -ro-stem, cf. Avestan tigra-‘sharp, pointed’ : tizi-sruua-‘having pointed horns’, Parmentier had already assembled the basic facts that later became known as the “Caland system”.

In the following decades interest in the s-stems has often been taken from this perspective, with scholars trying to add new suffixes to the Caland system on the one hand and to elucidate the prehistory of the system on the other: Where does the enigmatic -i- in the compounds come from and was the Caland system really just a mechanical rule of substitution of suffixes to switch between word classes? The last word on these questions does not seem to have been spoken, and M. is prudent enough not to burden his discussion of the s-stems with preconceived notions about their prehistory in terms of the Caland system.

The second chapter on neuter s-stems nouns in -os and -as deals with the question of their derivational basis (deverbal, deadjectival?), with their inflectional paradigm (proterokinetic and acrostatic types, according to Schindler (1975)) and their semantics. Several noteworthy results are achieved along the discussion of the material. 1. If ἐχθρός is to be derived from * ekstro-, the corresponding s-stem ἔχθος must have been built after the development of *kst > χθ, which is likely to be post-mycenaean, cf. myc. a 3 -ka-sa-ma /aiksmans/ vs. classical αἰχμή. In this case, ἔχθος, a “Caland” s-stem next to an adjective in -ro-, is clearly an inner-greek formation. 2. On the question of the inflectional paradigm of IE s-stems Schindler’s theory has been most influential, claiming that there was both a proterokinetic type CéC-s-0 (cf. Skt. kravis-, Greek κρέας, both from πιε krewh 2 -s-0; later remodeled to CéC-os-0), gen. CC-és-s (later remodeled to CeC-és-os) and an acrostatic type Ce:C-(o)s : CeC-es-(o)s. Firstly, while M. basically accepts Schindler’s proterokinetic type, he shows that many of the Greek examples taken to be instances of it are actually better explained within Greek itself. Thus in contrast to the epic form πένθος, πάθος is attested first in the tragedians and may be a back-formation from compound adjectives such as αἰνοπαθής, which is already Homeric. Secondly, M. rejects Schindler’s acrostatic type ( ἦθος : ἔθος, γῆρας : γέρας, etc.), but his arguments are less convincing here. In the case of ῥῆγος‘rug’ he assumes a Semitic loan (Arabic ruq’a‘piece of cloth’), for γῆρας he follows Frisk who assumed that the lengthened grade was taken over from the (s-?)aorist ἐγήρα. In the case of μήδεα‘plans, genitals’ : μέδεα‘genitals’ he refers to the s-aorist ἐμησάμην (Hom.+) which he tries to explain not from * med-, but from * meh 1 -, from which the present μήδομαι was built “for formal reasons and considering the close semantic relationship with μέδομαι“. But it is unclear if the root * meh 1 – is attested in Greek at all and why a new present should have been created from the aorist (cf. Skt. mmite, Av. fra-mima θ a). Furthermore, the gloss in Hesychius μῆστο· βουλεύσατο might point to an older athematic form * me:d-to which was remade into μήσατο. Finally, although in principle the Armenian a-stem mit(k’)‘thought’ may not go back to an IE s-stem, the easiest explanation seems to be that the frequent use of the plural * medesh 2 just as in Greek μήδεα gave Arm. * meteha > * mita which was reanalysed as an a-stem, cf. Matzinger (2005:47f.). It would seem, therefore, that at least in some formations the lengthened grade may indeed be inherited, although the distribution need not match the patterns set up by Schindler.

Since M. rejects a morphological explanation for Schindler’s second type, he gives a phonological explanation of the lengthened grade found in various languages, viz. a sort of “Szemerényi’s law” treatment in cases like * sed-s > * sess > * se:s (with long e), which is not verifiable. Furthermore, as he assumes analogical leveling of an original paradigm * sed-s : * sd-es- to * sed-s : * sed-es- followed by the development to * sess > * se:s (with long e), one wonders why in some cases the aberrant lengthened grade was retained while in other cases he assumes analogical shortening of the vowel in the nom./acc.

The following chapters dealing with inner-greek matters are more convincing and yield interesting results. M. argues that in Greek the derivational dependence of s-stem adjectives from s-stem nouns could be reversed in many cases, yielding numerous secondary s-stem nouns of no value for comparative reconstruction, e.g. in the case of φλέγος which is only attested in Hesychius and does not agree in ablaut with Skt. bhargas – ‘radiance’ and Lat. fulgur (from earlier fulgus) with which it has traditionally been compared, or in that of ἀληθής‘clear, true, unforgotten’ which is Homeric while the noun λᾶθος appears only in Theocritus. M. further assumes that many s-stem nouns may be deverbal, e.g. ἄημι‘to blow’ : δυσαής‘ill-blowing, unfavourable (wind)’ whence ἄος which is only attested in Hesychius. Many more examples for this type of situation can be given.

As for the semantics (ch. 2.6) of deadjectival s-stem neuters, M. relies on De Lamberterie (1989) who showed that while s-stems denoting graded qualities like τάχος‘speed’ may refer both to the positive end of the scale (‘high speed’) and to the phenomenon as such (‘speed’ in general, whence it does also make sense to talk of ‘slow speed’), the corresponding abstract nouns in – της only denote the former quality (‘high speed’) and are therefore less frequent than the s-stem neuters. As τάχος also covers the whole scale of ‘speed’ there is no need for a corresponding noun meaning ‘slowness’ with the same range of meaning. The correlative form of τάχος is therefore βραδύτης, with βράδος being late and rare. The consequence of this reasoning for the question of the “Caland system” is that a mechanical derivational rule “-u-stem adj. : s-stem noun” is too simplistic, as it does not take into account the semantics of the adjective.

The analysis of deverbal s-stem nouns is less fruitful as was the research done on this question before M. Thus, there is no obvious difference in meaning between δέος and δεῖμα both ‘fear’ or ἔσθος and εἷμα both ‘garment’. His basic point, though, that the distinction between, say, derivatives in – μα and those in -s- must depend on the root meaning itself, may be a good starting point for further investigation.

The third chapter explores the animate s-stem nouns of the type ἔρως m. and αἰδώς f. While masculine animate s-stems tend to become -t-stems within Greek as in the case of ἔρως, ἔρωτος, the feminines αἰδώς and ἠώς retain their archaic s-stem inflection, cf. gen. αἰδοῦς and ἠοῦς. The reason for this M. finds in an analogy to fem. nouns in – ω such as ἠχώ, ἠχοῦς and φειδώ, φειδοῦς that, just like the fem. s-stems, are sometimes personified (divine Ἠώς and Ἠχώ). Apart from the nom. and voc. sg. the paradigms of both inflectional classes looked alike.

As for the IE background of these formations, M. proposes a derivation different from the communis opinio that forms like ἠώς go back to neuter collectives in *- h 2 which as in the case of the thematic forms in * -e-h 2 were reanalyzed as feminines, maybe starting in the very word for ‘dawn’ (as proposed, e.g., by Fritz (1998)). M. rightly points out that there is no evidence for the supposed underlying neuter in *- os and that unlike the collectives in – or such as Greek ὕδωρ the forms in – os are never neuters, making a derivation from a preform in * -os-h 2 less likely. Therefore, his proposal to derive IE * h 2 ewso:s from * h 2 ewsos-s seems worth considering in future research.

The fourth chapter is concerned with s-stem adjectives, both simple (like ψευδής) and compounded (like δυσμενής). M. points out that not all s-stem adjectives are denominal, but seem to be derived from the corresponding verb, e.g. τηλεφανής‘visible in the distance’ from φαίνομαι‘appear, be visible’. Furthermore, as stressed already in the introduction, there are no word equations for simple s-stem adjectives. Both phenomena therefore seem to be Greek innovations.

As for the first point, in some cases an adj. in – ης seems to correlate to a “passive” aorist in – ην, cf. ἡμιδαής‘half-burnt’ beside ἐδάη‘(was) burnt’, thus it does not seem improbable that the adjective was in fact derived from the stem of the aorist, a process for which Armenian provides a good parallel in forms like miaynkeac’‘living alone’ from the aorist 1. sg. kec’i, 3. sg. ekeac’.

As for the second point, many simple s-stem adjectives are attested later than the corresponding compound forms, e.g. μιγής‘mixed’ (Nic. fr. 68.4) from earlier παμμιγής‘all-mixed’ (A.+) or ὀφελής‘advantageous’ ( P.Oxy. II 237.8.15) from ἀνωφελής‘useless’ (A.+) whence also the verb ὠφελέω‘be of use’ with initial ω‐ that is best explained as deriving from the compound form.

In sum, M.’s book may be seen as an exemplary work of the sound philological investigation that should always precede any attempt to reconstruct words and patterns for the proto-language. His detailed study of the formal and semantic aspects of the Greek s-stems is certainly a good step towards a more balanced picture of what is inherited and what is innovated. Let us hope that more word classes receive a similar treatment in the future.

Some minor notes:

p. 55: On the outcome of IE s-stems in Armenian: Not all s-stems have become o-stems, cf. e.g. the a-stem mit‘thought, idea’ mentioned above. Other s-stems have become i-stems in Armenian, e.g. the adj. barjraberj‘very high’ from a nom. in *- bherg’he:s (cf. Skt. – bárhas), which developed into – i(h), cf. Matzinger (2005:50).

p. 104: On the question of markedness of one pair of an opposition (type δεξιός, ἀριστερός) M. claims that ” ἄρσην does not even have a form with ‐τερο‐“, but in Mantineia (IG 5(2) 262.21) we find αρρεντερος. Of course, this does not invalidate his general point.

p. 183: ἐγχεσίμωρος is given as meaning ‘fighting with the spear, mad for spears’. Is the latter meaning given because of μωρός‘stupid’? But maybe there is a different root involved here, viz. ‘famous, renowned’, which is found in OIr. már‘big, renowned’, Welsh mawr‘id.’, Goth. wailamereis‘praiseworthy’ and PNN such as Gaulish Nerto-, Segomarus, OCS Vladimer- and OHG Valdemar. In this interpretation, ἐγχεσίμωρος can be put in line with Homeric ἰόμωρος (said of the Achaeans) and ὑλακόμωρος (said of dogs), cf. also Heubeck (1987:161f.).

pp. 203-205: M. does not give semantic parallels for his ingenious explanation of Greek ὑγιής‘healthy’ as *h 2 ugi-h 1 eh 1 s-‘sitting strong’. Furthermore, it is questionable if the root involved as a second member is to be reconstructed as *h 1 eh 1 s- (as is commonly done), since a priori one would not expect a verb with stative meaning to be reduplicated (one would rather expect the simple root in that case) and the Anatolian evidence speaks against a sequence *eh 1 in this case as in HLuv. isa-‘sit’ and istarta-‘throne’ must go back to a pre-form in * e:s-, since *eh 1 results in * a, cf. Lyc. tadi‘puts’ < IE * d h eh 1 – (: Hitt. tezzi‘says’), cf. Kimball (1999:122). It seems more likely to me that the verb ‘to sit’ was originally identical to ‘to be’, IE * h 1 es-. A compound with this root in the full grade as a second member should look something like * h 2 ugi-h 1 es in the masc./fem. form which does not lead us to the Greek form. Finally, under the assumption of a second element * h 1 eh 1 s -, the neuter in – ες would have to be explained analogically.


Ch. De Lamberterie, ‘Vitesse, rapidité, lenteur: fonctions suffixales en grec classique’, LALIES, Actes des sessions de linguistique et literature, vol. 7, 1989, pp. 275-277.

M. Fritz, ‘Die urindogermanischen s-Stämme und die Genese des dritten Genus’, in W. Meid (ed.), Sprache und Kultur der Indogermanen. Akten der X. Fachtagung der indogermanischen Gesellschaft, Innsbruck, 22.-28. September 1996. Innsbruck: Institut für Sprachwissenschaft, 1998. pp. 255-264.

A. Heubeck, ‘ Ἰόλαος und Verwandtes’, MSS 48, 1987, pp. 149-166.

S. E. Kimball, Hittite historical phonology. Innsbruck: Institut für Sprachwissenschaft, 1999.

J. Matzinger, Untersuchungen zum altarmenischen Nomen: die Flexion des Substantivs. Dettelbach: Röll, 2005.

L. Parmentier, Les substantifs et les adjectifs en -es- dans la langue d’Homère et d’Hésiode. Ghent and Paris: Vanderhaeghen, 1889.

J. Schindler, ‘Zum Ablaut der neutralen s-Stämme des Indogermanischen’, in H. Rix (ed.), Flexion und Wortbildung: Akten der V. Fachtagung der indogermanischen Gesellschaft. Wiesbaden: Reichert, 1975. pp. 259-267.

[For notes on this review by Michael F. Lane, please see BMCR 2007.10.17.]