Φωνήεντα Chapter 1

Introduction

by C Ryan Moniz

original research· spring 2010 - harvest 2016
updated & published· spring 2022

philology

Φωνήεντα·

1 » 2 · 3 · 4 · 5

Indo-European nominal ablaut

The articles in this series seek to examine the different approaches to accent-ablaut variation in Indo-European nouns, and what insight these approaches might provide for explaining the quality of vowels in Proto-Indo-European. They focus on the nouns that are called “athematic” — i.e. those nouns that lack a “thematic” vowel *-e/o- that remains in the inflectional ending — because “thematic” nouns typically yield non-ablauting nouns in the Indo-European daughter languages (Sanskrit a-stems, Germanic *a- and *ō- stems, 1st and 2nd declension nouns in Greek and Latin, etc.). Moreover, athematic nouns, with their alternating accent placement and changing vowel quality, are often presumed to be older, or even represent the original case system of PIE, with thematic nouns resulting from clause alignment shifts and subsequent regularization.❦1 Fortson 2004 p 87, 108, 157, 113·❦2 Matasović 2012 p 18·❦3 Kloekhorst 2008 p 137

The term “accent-ablaut” refers to a system of morphophonemic variation that includes vowel apophony (between *e*, *o, long *e:, and zero-grades) and movement of the accent. The term itself assumes an inherent interrelationship between the two phenomena. The accent-ablaut allomorphy manifests itself as a distinction between the “strong cases,” usually the nominative and accusative cases (though the Sanskrit accusative plural was not), and the “weak cases,” i.e. the genitive and other oblique cases.❦4 Halle 1997 p 291 The accent of the vocative case behaves strangely, especially in Sanskrit, but this was probably for phrase-level prosodic reasons, and because its ablaut parallels that of the nominative, it is classed as “strong” in PIE.❦5 Kiparsky 2010 p 10 Due to the relative archaism of this sort of allomorphy, there are limited examples of accent-ablaut extant within any but some of the most ancient Indo-European langauges; some rare examples extant within the same language are Sanskrit NOM.SG dā́ru ~ GEN.SG dróḥ ‘wood’ and Old Irish NOM.SG ben ~ GEN.SG mná ‘woman.’❦6 Klimp 2013 p 15 More typically, evidence for an ablauting noun is found in variation between the vowel grade of reflexes between different language, for example Old Norse tǫnn and Old English tóþ ‘tooth’ from Proto-Germanic NOM.SG *tanþs vs Gothic tundus ‘tooth’ from Proto-Germanic GEN.SG *tundiz.❦7 Ringe 2006 p 280 When evidence comes from separate languages, it is still desireable that an ablauting paradigm be reconstructible for the ancestor of at least one Indo-European subfamily (e.g. Proto-Germanic NOM.SG *tanþs ~ GEN.SG *tundiz).

History

While ablaut was of considerable interest for early Indo-European studies, most early Indo-Europeanists did not pay particular attention to the accent patterns of the language they were reconstructing, except perhaps to mention that, in contrast to many of the modern Indo-European languages, the accent was “free.”❦8 Sandell 2014 p 2 Brugmann (1930), for example, claims that Proto-Indo-European accent was free, and that by no recognizable external conditions is the accent bound to any particular place in the word (p 949). However, some linguist, such as Hirt (1985), did note a relationship between accent and ablaut in Indo-European reconstruction: “Der Ablaut dieser Sprachepoche wird unzweifelhaft mit Akzentunterschieden zusammenhängen” [The ablaut of this speech-period is certainly related to accent differences] (p 15).

The status of the ablaut-accent relationship in Indo-European studies changed radically with the works of Pedersen (1926) and Kuiper (1942). They identified the accent of athematic nouns as moving between three “parts” of the noun: (1) within the root, (2) within a derivational/noun-class suffix, & (3) before the inflectional case/number ending. Their analyses created distinction between what have been termed ”proterodynamic” and “hysterodynamic” patterns of athematic nominal ablaut.❦9 Kloekhorst 2013 p 1 Proterodynamic nouns exhibited alternation between an accented *e-grade root in strong cases, and an accented *e-grade derivational suffix in the weak cases. Hysterodynamic nouns had alternation between an accented *e-grade suffix in the strong cases, and an accented *e-grade inflectional ending in the weak cases.❦8 Sandell 2014 p 1

Table 1 — Pedersen & Kuiper’s model
Proterodynamic Hysterodynamic
Root ·Suffix -Ending R ·S -E
STRONG é ∅ ∅ ∅ é ∅
WEAK ∅ é ∅ ∅ ∅ é

A noteworthy consequence of this model is that, at least for the athematic nounse, the vowel placement is entirely predictable based on the placement of the accent, and vice versa — wherever there is no accent, there is no *e-grade vowel.

At the 1964 Erlanger Kolloiquium, scholars such as Rix, Schindler, Eichner, and Hoffmann discussed the works of Pedersen and Kuiper, and postulated a nominal accent-ablaut system which has since been dubbed the “Erlangen model.”❦9 Kloekhorst 2013 p 1 This model consists of four accent-ablaut patterns, two of which continue the previous proterodynamic and hysterodynamic (now termed “proterokinetic” and “hysterokinetic,” respectively), and the other two which were added: the “acrostatic” type, wherein the accent is fixed on the root in all cases with varying *o- or long *e:-grade in the strong cases and short *e-grade in the weak cases, and the “amphikinetic” type, which shows alternation between an accented *e-grade root and *o-grade in the suffix in the strong cases, and accented *e-grade on the inflectional ending in the weak-cases.❦10 Sandell 2014 p 1

Table 2 — Erlangen model
Acrostatic Proterokinetic Hysterokinetic Amphikinetic
R ·S -E R ·S -E R ·S -E R ·S -E
STRONG ó/éː ∅ ∅ é ∅ ∅ ∅ é ∅ é o ∅
WEAK é ∅ ∅ ∅ é ∅ ∅ ∅ é ∅ ∅ é

The amphikinetic type is particularly unique in that the strong cases have two full-grade vowels — an accented *e-grade in the root, and an unaccented *o-grade in the suffix — and because the accent alternates between the root and the inflectional ending, which are divided by the suffix morpheme.

One of Kuiper’s students, Beekes, had a different analysis of Kuiper and Pedersen’s work and, along with Kortlandt, developed what is now called the “Leiden model.”❦9 Kloekhorst 2013 p 1 This model only posits a ”static” pattern (formally identical to the Erlangen model’s acrostatic pattern) along with the original proterodynamic and hysterodynamic patterns, with the innovation that the nominative singular of the hysterodynamic pattern bears the accented *e-grade vowel, as opposed to the expected accent on the suffix, which is still the case for the accusative singular.❦10 Sandell 2014 p 1 With the exception of the static paradigm, the Leiden model restores a one-to-one correspondence between the accent and *e-grade of athematic nouns.

Table 3 — Leiden model
Static Proterodynamic Hysterodynamic
R ·S -E R ·S -E R ·S -E
NOM ó/éː ∅ ∅ é ∅ ∅ é ∅ ∅
ACC.SG ó/éː ∅ ∅ é ∅ ∅ ∅ é ∅
WEAK é ∅ ∅ ∅ é ∅ ∅ ∅ é

This model is not as widely cited as the Erlangen model; however, the Leiden and Erlangen models together are the most prominent examples of paradigmatic approaches to Indo-European nominal accent-ablaut.

An issue that has been raised against paradigmatic approaches is that they make no effort to explain the surface accent of Indo-European nouns, instead taking the paradigmatic templates as ”givens” of PIE grammar. in response to this deficiency, many Indo-Europeanists have turned to a compositional approach in order to explain the processes through which the observed allomorphy of Indo-European athematic nouns arises. Kiparsky, together with Halle, pushed forward the idea that accent (or lack thereof) was a feature inherent to morphemes in the protolanguage, much as they are in most analyses of Baltic and Slavic languages.❦11 Kiparsky & Halle 1977·❦12 Halle 1997·❦13 Kiparsky 2010 It was Kiparsky (1973) who developed the Basic Accentuation Principle (BAP), which erases all underlying accents but the leftmost underlying accent of a word, and places an accent on the leftmost syllable if no morphemes in the word are inherently accented.❦14 Kiparsky 1973 In recent years, compositional approaches to PIE accent-ablaut have received more attention with the introduction of Optimality Theory into Indo-European linguistics; the application of Prince & Smolensky’s (1993) theory of conflicting phonological constraints to PIE has provided new avenues for explaining the many accent patterns observed in athematic nouns. Despite recent advances in this vein, however, the paradigmatic models for PIE remain the most widely referenced explanations for the various accent-ablaut patterns.

Vowel phonemes in Proto-Indo-European

De Saussure was the first to suggest that the common late 19th century assumption of the Greek i e a o u model for Proto-Indo-European language might be false in his Mémoire sure le système primitif des voyelles dans les langues indoeuropéenes (1879). Firstly, de Saussure demonstrated that *i and *u were phonemically in a different category than the “true vowels,” and should more accurately be grouped with the “resonant” consonants *r, *l, *m, *n. The resonants and the high vowel phonemes corresponded in phonetic alternation between [–syllabic] and [+syllabic] allophones (*[r ~ r̩], *[l ~ l̩], *[m ~ m̩], *[n ~ n̩]; *[i̯ ~ i], *[u̯ ~ u]) and had parallel ablaut gradation, such as the *e ~ *o ~ ∅ demonstrable in Gk❦15 Pulleyblank 1965 p 87:

πένθος; πέπονθα ~ ἔπαθον
← *kʷéndʰ-os; *kʷé-kʷondʰ-h₂e ~ h₁é-kʷndʰ-om

λείπω ~ λέλοιπα ~ ἔλιπον
← *léikʷ-oh₂ ~ *lé-loikʷ-h₂e ~ h₁é-likʷ-om

More radical, however, was de Saussure’s proposals of an interaction between some consonant phoneme and *e which distinguished theretofore reconstructed *a and *o. De Saussure’s hypothetical consonant was later identified with Möller’s ”laryngeal” *h₂ (probably *[ħ/χ]) attested in Hittite ḫ and other Anatolian languages; two other ”laryngeals,’ *h₁ and h₃ (probably *[ʔ] and *[ʕ/ʁ], respectively) have also been reconstructed.❦16 Pulleyblank 1965 p 89·❦17 Simon 2010 This theory of for the origin of an *o ~ *[a] distinction is now generally accepted, or at least acknowledged to some extent, as it allows for an explanation of ablaut in words with previously reconstructed *a.

On the basis of the Anatolian evidence the existence of a PIE phoneme *a is doubtful. As noted by Lubotsky (1989), it “has limited occurrence, is absent from endings and suffixes, shows practically no ablaut, and is confined to a few isolated words not belonging to the so-called ‘vocabulaire fondamental’” (p 53). Most of the instances of a in those Indo-European languages where it was not in some way conflated with *o — of which, Lubotsky claims, only Greek is consistently reliable — can be explained by reflexes of the reconstructed ‘laryngeal’ *h₂. Despite the near consensus on the issue at this point in time, there are enough inconsistencies between reflexes of *a in Indo-European languages for reconstructions to still include it as a phoneme in certain instances. However, Lubotsky found a way to account for the inconsistencies between these reflexes using the advances that had been made in understanding Indo-European accent and ablaut. Especially on the basis of Beeke’s analysis of accent-ablaut (1985) and syllable structure (1988), Lubotsky explained that the reflexes in nouns such as

⠀Greek ἅλς, Latin sāl, Old Church Slavonic solь ‘salt’ and
⠀Greek χήν, Latin ānser, Sanskrit haṃsáḥ ‘swan’

can be explained by the hysterodynamic paradigm of suffixed nouns

⠀← *séh₂·l-s NOM.SG ~ *sh₂·él-m ACC.SG ~ *sh₂·l-és GEN.SG and
⠀← *gʰéh₂n·s-s NOM.SG ~ *gʰh₂n·és-m ACC.SG ~ *gʰh₂n·s-és GEN.SG, respectively

and that the Latin nāris, Sanskrit nā́sā, Old Church Slavonic nosь ‘nose’ can similarly be explained with proterodynamic

⠀← *néh₂·s-s NOM.SG ~ *néh·s-m ACC.SG ~ *nh₂·s-és GEN.SG.❦18 Lubotsky 1989 p 60

The accent-ablaut patterns proposed by the Leiden model were instrumental in Lubotsky’s final elimination of PIE *a as anything other than simply an allophone of the phoneme */e/ next to *h₂.

Given the absence of a phonemic *a, only mid-vowels *e and *o (and their long counterparts) are reconstructible for PIE. The typologically peculariar absence of a phonetically low vowel with phonemic status in the Indo-European protolanguage should be noted. Chapter 5 features further discussion of this issue.

Φωνήεντα·

1 » 2 · 3 · 4 · 5

references

❦1· Fortson 2004 p 87, 108, 157, 113
❦2· Matasović 2012 p 18
❦3· Kloekhorst 2008 p 137
❦4· Halle 1997 p 291
❦5· Kiparsky 2010 p 10
❦6· Klimp 2013 p 15
❦7· Ringe 2006 p 280
❦8· Sandell 2014 p 2
❦9· Kloekhorst 2013 p 1
❦10· Sandell 2014 p 1
❦11· Kiparsky & Halle 1977
❦12· Halle 1997
❦13· Kiparsky 2010
❦14· Kiparsky 1973
❦15· Pulleyblank 1965 p 87
❦16· " p 89
❦17· Simon 2010
❦18· Lubotsky 1989 p 60

bibliography

philology