Response by

In a recent review (BMCR 2003.06.31) Dag Haug raised a number of severe objections regarding my book “Die Sprachform der homerischen Epen” [hereinafter SHE] (Wiesbaden: Dr. Ludwig Reichert Verlag 2002). Since Haug’s account grossly misrepresents what I wrote (repeatedly overgeneralizing his criticism and addressing only a fraction of the evidence), and is also overly-polemical (e.g., his accusation of verbosity [see below I.3] creates the impression that my points are not worth discussing), Haug’s remarks call for some comments on my part and the rather unpleasant task of reviewing a review.

At the outset Haug remarks that “most of the book contains technical linguistic discussion which will deter many classicists” — an unfortunate comment that could be taken to imply that a combination of technical linguistic discussion and the Classics might be a deterrent in general and hence should be avoided. Surely Haug would not subscribe to this view himself, for as he admits a few lines later, “the field of Homeric linguistics demands a symbiosis of classical philology and Indo-European linguistics”. Are we to conclude that Haug is advocating a less technical approach to Indo-European linguistics? I agree that it is difficult for uninitiated classicists to follow highly technical analysis in Indo-European linguistics; being fully aware of this, I had explicitly pointed out in my preface (pp. xv-xvi) that the various chapters would be likely to interest different readers. Despite such interdisciplinary hurdles, however, a thorough Indo-European analysis based on the maximum relevant information continues to be indispensable.1

I. Programmatic issues

I.1. Myceneanisms

In his first paragraph Haug notes that “H’s book addresses this question too and answers it in the negative, which is hardly surprising in view of the author’s intention to emphasize the importance of recent Ionic elements.” Haug insinuates that my negative assessment of the hotly-debated Myceneanisms has to do with an intention to emphasize Neo-Ionic. But why should the presence of Neo-Ionic forms preclude Myceneanisms? The connection drawn by Haug would be as monstrous as it is illogical; in any case, there is not a single line in my book to attest to such a connection.

I.2. Anachronisms

The notion of “anachronism” is central to my investigation, hence its appearance in my subtitle and in the table of contents. In fact, I made an attempt to exploit the significance and potential of such anachronisms to reflect traits of contemporary vernacular. Positing an Ionic vernacular is not my invention but goes back at least to Witte and Parry (SHE pp. 22f). The unsolved problem so far has been how to obtain evidence for the Ionic vernacular, given the scarcity or absence of contemporary sources at an early stage of Greek literacy. I sought a partial answer to this core problem by pointing out the significance of anachronisms as indirect evidence for such vernacular forms.

The likelihood that anachronisms might display vernacular forms is borne out by typological observations, see chapter III pp. 46-77. These stress the interplay of extra-linguistic factors and language change and show in particular that historical and cultural changes (including the beginning of literacy) might result in a freer (unstigmatized) use of vernacular varieties, see pp. 57-60, and especially pp. 69-77. Haug raises two objections against my conception of “anachronisms”:

a) “Hackstein most often compares ‘young forms’ in Homer with their Attic equivalents.” This is not so: after all, Attic might be more conservative than Ionic. In determining anachronisms I dealt with coexisting variants within the Homeric language (cf. SHE p. 24 ἱππεῖς / ἱππῆες, pp. 26f. ὅτις / ὅστις, pp. 32f. ἄσβεστος / ἄσβετος). Outside the Homeric corpus I frequently chose as a point of reference later literary and dialectal Ionic, cf. e.g. pp. 98-101, with a systematic conspectus of thematized mi-presents in literary Ionic.

b) Haug generally dismisses the significance of anachronisms by concluding that “It seems unnecessary to make so much fuss about ‘anticipatory anachronisms’ in treating this well-known phenomenon”. Anachronisms are indeed a well-known phenomenon, but this does not diminish their importance or reduce the need for thorough investigation, especially in light of the methodological fallacies (which Haug does not mention in this context) set out at length in my book (pp. 19-22, 78-80), such as the misconception that older strata of languages demand older forms and, therefore, the emendation of anachronistically young forms.

The two notions of anticipatory and isolative (i.e. abortive) anachronism (my own distinction) are central to what follows, since I discuss large-scale language drift phenomena in the following chapters. Among these we find the Ionic thematization, the spread of alphathematic inflection, new ablaut patterns, and the connection of the pluperfect of οἶδα with athematic preterites (imperfects/aorists).

I.3. Haug’s account of the notion of anachronism is thus lacking in sophistication, and Haug is off the mark in stating that “Verbosity is a general problem with this book”. For this he cites my discussion of προθέουσιν A 291 (pp. 112-117) as an example, claiming that “the five pages on προθέουσιν… bring nothing new compared to what Ameis-Hentze said on the form, except for banalities like …”. At the same time, he refrains from discussing any of the points I actually made. The reader of Haug’s review is once again left with a misleading (and, I would say, inaccurate) judgement. There is no need to repeat my discussion of the form here. Suffice it to say that I began with a sketch of the history of the problem, which seemed appropriate in light of the great number of explanations proposed over the years. Then, and despite Haug’s “nothing new”, I suggested a new solution, showing that a likely explanation of προθέουσιν emerges if the form is viewed as part of a word game (for which I adduce several Homeric examples), and if it is subject to Ionic thematization, and finally if one bears in mind the stylistic traits of actor language. One may, with Haug, call this a banality, but not if the phenomenon of narrator versus actor language can plausibly be shown to account for the (artificial) mutilation of the underlying form * προτιθεῖσιν in a highly emotional context (the confrontation between Akhilleus and Agamemnon).

I.4. Aeolicisms

Haug seems to think I am overly skeptical of Aeolicisms and that I may have underused this option, but I fail to see his point. Why should a form be Aeolic, if it could be Ionic or simply an archaism? I still think that Witte’s and Parry’s taxonomy (cf. pp. 39f) and the priority of Ionic are essentially valid — not to imply that the presence of Aeolicisms in the Homeric corpus is called into question: this is undisputed. Nevertheless, the question whether a particular form might be an Aeolicism (as opposed to being explained as Ionic or as an archaism) has to be examined anew in each case. Thus, for instance, is θεά in the opening line of the Iliad an Aeolicism (as commonly believed)? Here, Martin Peters has pointed out the occurrence of θεά in archaic West Ionic graffiti (see M. Peters in L. Isebaert (ed.), Miscellanea Graeco-Latina, Namur 1993 p. 93 fn. 27 and Antonín Bartonek & Giorgio Buchner, Die Sprache 37,2 (1995), p. 194), suggesting that θεά might be West-Ionic, cf. also SHE pp. 197ff.

I.5. “Kunstsprache”

Haug’s remark that Hackstein “seems to underestimate the force of the Kunstsprache in his eagerness to show that traditional Homeric philology has overestimated it” ignores the many passages where I discuss “Kunstsprache/Dichtersprache” (e.g. SHE pp. 36-38, 39ff, 46, 89f, 115f, 119, 167, 195f, not to mention my articles devoted to this very issue). Certainly the subject index (pp. 305-308) makes clear that underestimating the force of the “Kunstsprache” is not on my agenda. (It is baffling how Haug can maintain this while at the same time refusing to accept “Kunstsprache” as part of the explanation of the “mystery form” προθέουσιν cf. I.3 above.)

I. 6. The pluperfect of οἶδα

Haug criticizes my proposal to attribute the vacillation between ‐ει and ‐η found in the 3rd sg. pluperfect and in the case of Homeric ἔσβη for ἔσβει to a hypercorrection triggered by the same vacillation in the imperfect of athematic mi-verbs, e.g. ἐτίθ and younger ἐτίθ. He maintains that the envisaged conversion of ‐ει into ‐η is contradicted by the converse development found in the preterite of mi-verbs to roots in ‐η (*eh1), namely ἐτίθη —> ἐτίθει, ἄη —> ἄει (SHE p. 119).

The alleged contradiction, however, is a pseudo-problem. It is typical for a hypercorrection to reverse a diachronic process that otherwise surfaces synchronically as a random vacillation: cf. SHE p. 262, where I explicitly speak of a “gegenläufige Entwicklung”. Neither the presumed hypercorrection of ᾔδει to ᾔδη nor that of ἔσβει to ἔσβη is invalidated. Haug’s conclusion that “there seem to be no parallel cases for the supposed hypercorrection of ᾔδει to ᾔδη” is simply false. Also note that there are other firm indications of a mutual attraction of the paradigms of ᾔδεα and ἐτίθην : cf. Hdt. ὑπερτίθεα modeled after ᾔδεα, and conversely εἰδείην after τιθείην and so forth, see SHE pp. 262f. Analogy need not be monodirectional.

Haug goes on to criticize my proposal to identify the striking ablaut pattern of 3sg. ᾔδει (*weid-) versus 3. pl. ἴσαν (*wid-) with that of the root aorist. He distorts my argument by connecting two things which I did not, namely, on the one hand, the ablaut pattern of ἔβη : ἔβαν, which is secondary (SHE p. 258 fn. 5), and on the other that of ἠείδει : ἴσαν. As I clearly stated, the paradigm ἠείδει : ἴσαν in terms of ablaut does not represent a secondary ablaut relationship (like ἔβη : ἔβαν), but the primary and inherited formation. The “preterite of οἶδα” (as isolated as it stands compared to the other type of pluperfect, e.g. πεποίθεα) displays an archaism that finds a perfect match in Vedic, i.e. 3sg. ávet : 3pl. ávidur (SHE p. 259). Haug withholds this part of my argument from the reader.

II. Haug’s remarks on individual forms

II.1. ad SHE p. 99 and 119: Haug’s ex cathedra judgement “The case for restoring κνεε… is not convincing” ignores the preference for dactyls in the fifth foot, observed already by R. Meister, Homerische Kunstsprache 72, a factor so strong as to demand the substitution of ‐ως by ‐μενος in perfect participles, see O. Hackstein, Glotta 74 (1997/98), pp. 41f. As for Haug’s suspicion that “The author seems unaware that later Ionic has κνα”: later Ionic κνα does not preclude the scenario I suggested on p. 119, since Aristarchus’s κνέε could be generated by positing a thematized * κνῆε that underwent hiatus shortening (as per SHE p. 160), cf. likewise M. L. West, JHS 118 (1998), 190, who however prefers to view the form as an Atticism (ibid. p. 191). I would be inclined to think of the transmitted κνῆ as the original form (i.e. an athematic imperfect3) and κνέε as a Neo-Ionic by-form, which would be preferred in the thesis of the fifth dactyl.

II.2. ad SHE pp. 103ff: Against my proposal to explain Ionic εἰς”you are” as a thematized form Haug objects: “since it [Att. εἶ < *ehi < *h1e(s)si] is an archaism, Attic must have inherited it from the common Attic-Ionic dialect, which means that it must have existed in the prehistory of Ionic.” But even the presence of * εἶ in Old Ionic would not preclude a subsequent thematization; after all, the tendency of the Homeric (Ionic) paradigm of εἰμί to be thematized is well-known. Thus, beside the inherited optative Ionic (= Attic) εἴη we find for instance secondary Ionic ἔοι Il. 9.142, 11.838, etc. The explanation I suggested for εἰς can hardly be ruled out: it has the advantage of explaining the peculiar accentuation pattern of εἰς (see SHE pp. 104f) and of connecting εἰς to a phenomenon (thematization) otherwise well established for Homeric presents, including the paradigm of εἰμί itself.

II.3. ad SHE p. 132: Haug dismisses the possibility that Sappho’s ἔσσο (1,28 PLF) might be an epicism; but the interrelationship between Sappho’s literary dialect and epic language continues to hold (albeit to varying degrees) and applies to Sappho 1 PLF as well, see E. Tzamali’s commentary (Syntax und Stil bei Sappho, Dettelbach: Röll Verlag 1996, pp. 38-90), and cf. line 7 ἔκλυες, line 10 ὤκεες.

II.4. ad SHE p. 139: As for ἐγκάτθετο Il. 14.223, Haug is right to point out that the apocope in the preverb κατ might reflect an Aeolicism. He concludes: “If so, it [ ἐγκάτθετο ] must be archaic Aeolic, for in historical times the kappa forms have intruded at least into the 3rd plural.” First, it is improper to infer from one possibly Aeolic morpheme that the entire form must be Aeolic, as this underestimates the potential of the Kunstsprache to create Mischformen (on which see SHE p. 37, 66f), cf. also Od. 11.74 κακ. Besides, metrical considerations (avoidance of unmetrical * ἐγκατάθετο) may have played a role, cf. P. Chantraine, Grammaire homérique I p. 87. Furthermore, the spread of the k-morpheme is not restricted to Aeolic, but extends over several Greek dialects including Ionic, see SHE p. 137 with references.

II.5. ad SHE pp. 105f [sic, i.e. 108f]: As for my proposal to derive Cretan πρεῖγυς from *preti-, Haug correctly points out that Kiparsky and Cowgill restricted the sound change of *-eti > ‐ει to final syllables, as I noted myself on p. 108 (“vor wortauslautendem -i”). However, it remains perfectly reasonable to reckon with an analogical extension of an erstwhile preposition * πρετι > πρει as a bound preverb, cf. the generalization of προς (< prevocalic *proty) as a preverb. This scenario is rendered more likely by the occurrence of ποι as the first member of compounds, such as Delphic ποιτασσόμενον, ποικεφάλαιον. In these cases, the preverb ποι is most likely an analogically extended Kiparsky/Cowgill-variant deriving from *poti, as in Delphic ποὶ γᾶν.4

II.6. ad Hackstein SHE p. 156 note 12: Haug rejects the idea that a laryngeal was lost as the second of four contiguous consonants in internal syllables (i.e. πιε as posited in Historische Sprachforschung 115 [2002], 1-22 on the strength of numerous examples, passed over by Haug in silence). He submits that * dhro- and * tro- derivatives of set-roots preserve the laryngeal, but the picture is by far not as unequivocal as Haug would have us believe: one has to consider the possibility of thematic formants * -e-dhro-, * -e-tro-, cf. SHE p. 226 on βέρεθρον; moreover, he withholds the concluding remarks on p. 19 (op. cit.) where (explicitly addressing tro-formations) I hinted at the possibility of analogy, referring to Latin aratrum, which owes its medial long vowel to the influence of arare. As for Armenian arawr, Haug considers an analogical restoration of the laryngeal “impossible”, because (1) the suffix -tro- is not productive in Armenian, and (2) there is no corresponding verb * aram. Since, however, the rule posited (*-CHCC- > -CCC-) is of Proto-Indo-European date,5 one has to reckon with early restorations as well, which reduces the significance of the loss of a verbal paradigm of * ara- in Proto-Armenian.

II.7. ad SHE pp. 164ff: I supported Wackernagel’s claim that συνοχωκότε is ultimately linked to an adjective σύνοχος by pointing out the possibility of denominal adjectival formations being made into nouns without any intermediary verb (by analogy to participles of denominal verbs). I adduced typological parallels such as German gestreift, gebluemt, gemustert formed to Streifen, Blume and Muster (so called “Scheinpartizipien”6). Haug does not consider these sufficient to support my proposal, and goes on to claim that I fail to “give any Greek examples”. It is disconcerting, however, to see that he does not mention those Greek examples noted on p. 166, including footnote 21 with Homeric ἀρή —> ἀρημένος.

II.8. ad SHE pp. 246ff: Concerning my account of Homeric πέποσθε and the proposal that the anomalous 2 pl. perf. act. ending ‐σθε could be the generalised allomorph of an ending -σ [ :: epic 2nd sg. -σ ], Haug raises the objection that “it seems unreasonable that Greek should have gone through so many analogical steps to create an active ending which would be identical to the medio-passive ending.” Labelling the assumption of a 2nd pl. variant -σ (cf. ‐ς, ‐ς‐το, ‐ς‐ται) and the spread of an allomorph as “so many analogical steps” seems an exaggeration. Haug refrains from discussing the reasons that led me to prefer this solution over the others proposed (see SHE pp. 252f). Furthermore, Haug discounts παρθένος as evidence for the aspiration rule *-rst- > *-rht- > -ρθ because of the attestation of the same word in Doric dialects (e.g. Laconian πάρσενος), which would rule out underlying syllabic r. This leads him to reject Klingenschmitt’s etymology (otherwise accepted by LfgrE III 981 and Mayrhofer KEWA III 511 and 802). However, the aspiration rule is an inner-Greek rule7 postdating the desyllabification of syllabic *r, so that παρθένος remains valid as an example. Klingenschmitt’s derivation works for Homeric (Ionic) παρθένος, as it would for Doric παρθένος — whatever explanation of παρ one opts for. As for the question of the dialectal reflexes of syllabic *r, the picture is not as simple as Haug suggests, cf. I. Hajnal, Minos Suppl. 14 (1997), pp. 145ff.

III. Summary

In sum, Haug’s account of my book abounds in pontification based on ill-founded assumptions. Again and again, he misleadingly directs his criticism only to parts of my argument, passing over germane and even decisive points (see I.2, I. 3, I.5, I.6, II.2, II.6, II.7 above); therefore, one can only be disappointed with the way in which Haug has discharged his duties as a reviewer.

[[For a response to this response by Dag Haug, please see BMCR 2004.03.44.]]


1. A recent paradigmatic example of this sort, exemplifying the desired combination of Classics and an in-depth Indo-European treatment, is Alan J. Nussbaum’s Two studies in Greek and Homeric linguistics (Göttingen: Vandenhoeck & Ruprecht 1998).

2. Cf. recently M. L. West, Glotta 77 (2002), 125.

3. As for the accent, cf. Homeric στῆ and E. Schwyzer, Griechische Grammatik I pp. 377f.

4. The connection of ποἱ) and ποτἱ) has been known for some time, see V. Lüttel’s dissertation, κάς und καί. Göttingen 1981: Vandenhoeck & Ruprecht. 24f, 47, 78f. As for the derivation of ποι from ποτι, Lüttel (op. cit. 52, following P. Kretschmer, Glotta 1, 1909, 41ff) suggests an alternative scenario (dissimilatory loss of intervocalic -t- before a word with dental onset), but this can hardly be confirmed, see M. Peters, Die Sprache 27/2, 1981, 231f (No. 361). By contrast, it is much easier to analyze the form in the context of the examples discussed by Kiparsky and Cowgill.

5. See Historische Sprachforschung 115, 1ff passim, and cf. the early assimilation of PIE *-g-t- > *-k-t- as implied by -s- in Armenian dowstr.

6. Cf. Duden Bd 4: Grammatik der deutschen Gegenwartssprache. 5. Aufl. Mannheim 1985. p. 534, section 971.

7. Cf. H. Rix, Historische Grammatik des Griechischen. Darmstadt 1992. p. 78.