FACULTY OF HUMANITIES

UNIVERSITY OF COPENHAGEN

Ph.D. thesis
Bjarne Simmelkjær Sandgaard Hansen

Archaisms and innovations

four interconnected studies on Germanic historical
phonology and morphology

i
Contents
LIST OF ABBREVIATIONS ............................................................................................... V

1. Grammatical terms ....................................................................................................................................................... v

2. Linguanyms .................................................................................................................................................................. vi

1. INTRODUCTION ............................................................................................................. 5

1.1. Archaisms and innovations ....................................................................................................................................... 6

1.2. Aim and purpose of the present thesis ..................................................................................................................... 6

1.3. Structure and limitations of the present thesis ........................................................................................................ 7

1.4. General features of the individual articles ............................................................................................................... 8

2. METHODOLOGICAL CONSIDERATIONS ................................................................... 11

2.1. The Comparative Method ....................................................................................................................................... 11

2.1.1. How to apply the Comparative Method on linguistic data ................................................................................. 11
2.1.2. Potential shortcomings of the Comparative Method .......................................................................................... 12

2.2. Internal Reconstruction .......................................................................................................................................... 13

2.3. Analogy ..................................................................................................................................................................... 14

2.3.1. Proportional analogy .......................................................................................................................................... 14
2.3.2. Analogical or paradigmatic levelling ................................................................................................................. 14
2.3.3. Analogical processes of particular relevance to the present thesis ..................................................................... 15

2.4. Lexical borrowings .................................................................................................................................................. 17

2.4.1. Lexical borrowings from known sources ........................................................................................................... 17
2.4.2. Lexical borrowings from unknown sources ....................................................................................................... 17
2.4.3. Borrowing processes of particular relevance to the present thesis ..................................................................... 19

ARTICLE NO. 1: LAYERS OF ROOT NOUNS IN GERMANIC: CHRONOLOGY,

STRUCTURE AND ORIGIN .............................................................................................. 20

1. Communis opinio on root nouns ................................................................................................................................ 20

2. Material ....................................................................................................................................................................... 22

3. Three layers of Germanic root nouns ....................................................................................................................... 39

3.1. Layer I: Root nouns inherited from Proto-Indo-European .................................................................................... 39
3.1.1. Apparent counterexamples ................................................................................................................................. 40
3.2. Layer IIa: Unsegmentable substrate or loan words reanalysed as root nouns in Proto-Germanic ........................ 43
3.3. Layer IIb: Nouns from other declensions reanalysed as root nouns in Proto-Germanic or in parts of the Germanic
dialect continuum ......................................................................................................................................................... 44
3.4. Layer III: Nouns from other declensions reanalysed as root nouns in North Germanic ....................................... 45

4. Conclusion ................................................................................................................................................................... 47

ii
ARTICLE NO. 2: THE STRUCTURE, FORM AND FUNCTION OF THE GERMANIC
PRIMARY I-STEMS ........................................................................................................... 51

1. General characteristics of i-stems and their inflection ............................................................................................ 51

2. Form and function of the Germanic primary i-stems.............................................................................................. 54

2.1. Masculine i-stem verbal abstracts derivationally matching the ablaut grade found in the stem of the preterite
participle of a corresponding strong verb ..................................................................................................................... 54
2.1.1. I-stem verbal abstracts corresponding to class I strong verbs ............................................................................ 55
2.1.2. I-stem verbal abstracts corresponding to class II strong verbs ........................................................................... 62
2.1.3. I-stem verbal abstracts corresponding to class III strong verbs ......................................................................... 69
2.1.4. I-stem verbal abstracts corresponding to class IV strong verbs ......................................................................... 78
2.1.5. I-stem verbal abstracts corresponding to class V strong verbs ........................................................................... 81
2.1.6. I-stem verbal abstracts corresponding to class VI strong verbs ......................................................................... 82
2.1.7. I-stem verbal abstracts corresponding to class I-VII reduplicated strong verbs ................................................. 86
2.2. Masculine i-stem verbal abstracts corresponding to a strong verb but displaying unexpected radical ablaut grade
..................................................................................................................................................................................... 88
2.2.1. I-stem verbal abstracts corresponding to class I strong verbs ............................................................................ 89
2.2.2. I-stem verbal abstracts corresponding to class II strong verbs ........................................................................... 89
2.2.3. I-stem verbal abstracts corresponding to class III strong verbs ......................................................................... 92
2.2.4. I-stem verbal abstracts corresponding to class IV strong verbs ......................................................................... 97
2.2.5. I-stem verbal abstracts corresponding to class V strong verbs ........................................................................... 97
2.2.6. I-stem verbal abstracts corresponding to class VI strong verbs ......................................................................... 98
2.2.7. I-stem verbal abstracts corresponding to class I-VII reduplicated strong verbs ................................................. 98
2.3. I-stem nouns with no correspondence to a strong verb ......................................................................................... 99
2.4. I-stem adjectives ................................................................................................................................................. 105

3. Form and function of Indo-European primary i-stems ......................................................................................... 106

3.1. Indo-Iranian (represented here by Sanskrit) ........................................................................................................ 106
3.2. Greek................................................................................................................................................................... 108
3.3. The remaining branches ...................................................................................................................................... 110
3.4. Proto-Indo-European ........................................................................................................................................... 110
3.4.1. General types of i-stems in Proto-Indo-European ............................................................................................ 111
3.4.2. Derivational history of primary i-stems with radical zero grade ...................................................................... 111
3.4.3. Derivational history of primary i-stems with radical ο-grade .......................................................................... 112

4. Primary i-stems as a parallel type of the PIE toga/fuga formations ..................................................................... 113
4.1. I-stems of the toga/fuga type............................................................................................................................... 114
4.2. I-stems resisting analysis within the toga/fuga system ....................................................................................... 120

5. Conclusion ................................................................................................................................................................. 121

ARTICLE NO. 3: THE OUTCOME OF PIE *#HI- AND *#HU- IN GERMANIC ................ 122

1. PIE *#Hi- and *#Hu-: the classical view ................................................................................................................. 122

2. The development of PIE *#Hi- and *#Hu- revisited .............................................................................................. 123

3. Possible outcomes in Germanic ............................................................................................................................... 125

3.1. Material with an initial laryngeal followed by PIE *(V)i or *(V)u ...................................................................... 127
3.1.1. Possibility of PIE *#h1i- ................................................................................................................................... 127
3.1.2. Possibility of PIE *#h2i- ................................................................................................................................... 133
3.1.3. Possibility of PIE *#h3i- ................................................................................................................................... 143
3.1.4. Possibility of PIE *#Hi- (i.e. undeterminable timbre of the laryngeal) ............................................................ 143
3.1.5. Possibility of PIE *#h1u- .................................................................................................................................. 144
3.1.6. Possibility of PIE *#h2u- .................................................................................................................................. 148

iii
3.1.7. Possibility of PIE *#h3u- .................................................................................................................................. 155
3.1.8. Possibility of PIE *#Hu- (i.e. undeterminable timbre of the laryngeal) ........................................................... 155
3.2. Material with PG *#i- and *#u- reflecting the zero grade of PIE *# - and *# - ....................................... 158
3.3. Material with PG *#i- reflecting PIE *e /_NC .................................................................................................... 158
3.4. Material with PG *#u- reflecting the supporting vowel of PG *#uR- < PIE * - .............................................. 159
3.5. Material from other sources (analogical reshapings, onomatopoeias, lexical borrowings) ................................. 159

4. Ordering of data and preliminary conclusion ........................................................................................................ 160

5. Excursus: PG *ubila- ‘evil, bad’ and *ufna- ‘oven’ – why not †aubila- and †aufna-? ....................................... 168

6. Conclusion ................................................................................................................................................................. 170

ARTICLE NO. 4: THE OUTCOME OF PIE *- AND *- IN GERMANIC....... 171

1. The importance of the Germanic “Auslautgesetze” .............................................................................................. 171

2. The Proto-Indo-European state of affairs .............................................................................................................. 172

3. Apparent suffixal o-grade forms in the individual branches ................................................................................ 176

4. Accounting for the Germanic evidence: a selection of previous attempts ........................................................... 179

5. Accounting for the Germanic evidence: a new attempt ........................................................................................ 181

5.1. Remarks on the i-stem gen.sg. ............................................................................................................................ 184
5.2. Remarks on the i- and u-stem dat./loc.sg. ........................................................................................................... 185
5.3. Remarks on the u-stem nom.pl............................................................................................................................ 186
5.4. Diphthongal desinences reflecting PIE *- *- and *- *- or similar constellations .................................. 187

6. Conclusion ................................................................................................................................................................. 189

3. CONCLUSION ............................................................................................................. 191

3.1. Archaisms and innovations in nominal derivational morphology ..................................................................... 191

3.2. Support from Germanic historical phonology .................................................................................................... 192

REFERENCES ................................................................................................................ 194

ABSTRACT (ENGLISH) .................................................................................................. 209

ABSTRACT (DANISH) .................................................................................................... 210

iv
List of abbreviations

1. Grammatical terms
# word boundary N any nasal
abl. ablative n. neuter
acc. accusative nom. nominative
adj. adjective num. numeral
adv. adverb opt. optative
aor. aorist P any labial plosive
C any consonant pass. passive voice
c. commune perf. perfect
comp. comparative pers. personal
conj. conjunction pp.v. preterite-present verb
dat. dative prep. preposition
dial. dialect(al) pres. present
du. dual pret. preterite
f. feminine pron. pronoun
fut. future ptc. participle
gen. genitive R any resonant
H any laryngeal s.v. strong verb
indecl. indeclinable sb. noun of unspecified gender
inf. infinitive sg. singular
instr. instrumental T any plosive; any dental plosive;
intr. intransitive any obstruent
ipv. imperative tr. transitive
K any velar plosive any approximant
loc. locative V any vowel
m. masculine v. verb
mar. maritime voc. vocative
mid. middle voice w.v. weak verb

v
2. Linguanyms
Aeol. Aeolic Ital. Italian
Alb. Albanian Jutl. Jutlandish
Am. American Lat. Latin
Arm. Armenian Latv. Latvian
Att. Attic LG Low German
Av. Avestan Lith. Lithuanian
Boeot. Boeotian Lux. Luxembourgish
Burg. Burgundish Marruc. Marrucinian
Celt. Celtic ME Middle English
CLuw. Cuneiform Luwian MHG Middle High German
Crim. Crimean MIr. Middle Irish
Cz. Czech MLG Middle Low German
Da. Danish MW Middle Welsh
Dor. Doric Myc. Mycenaean
EFris. East Frisian NNorw. New Norwegian
EN East Norse Norw. Norwegian
Eng. English NWG North-West Germanic
Far. Faroese OCS Old Church Slavonic
Fun. (dialect of) Funen ODa. Old Danish
G German OE Old English
Gallo-Lat. Gallo-Latin OEN Old East Norse
Gaul. Gaulish OFrench Old French
Goth. Gothic OFris. Old Frisian
Gr. Greek OGutn. Old Gutnish
Gutn. Gutnish OHG Old High German
Hes. Hesych(ius) OIr. Old Irish
Hitt. Hittite OLat. Old Latin
HLuw. Hieroglyphic Luwian OLith. Old Lithuanian
Hom. Homeric ON Old Norse
Icel. Icelandic OPers. Old Persian
Ion. Ionic OPhryg. Old Phrygian

vi
OPr. Old Prussian Prov. Provençal
ORu. Old Russian PSl. Proto-Slavic
OS Old Saxon PWG Proto-West-Germanic
Osc. Oscan RN Runic Norse (Proto-Norse)
OSw. Old Swedish Ru. Russian
OW Old Welsh SCr. Serbo-Croatian
Pal. Palaic Skt. Sanskrit, Vedic
PAlb. Proto-Albanian Sw. Swedish
PCelt. Proto-Celtic Swi. Swiss
Pers. Persian Thrac. Thracian
PG Proto-Germanic Toch. Tocharian
Phryg. Phrygian Ukr. Ukrainian
PIE Proto-Indo-European Umbr. Umbrian
PIIr. Proto-Indo-Iranian W Welsh
PIt. Proto-Italic WG West Germanic
Port. Portuguese

vii
1. Introduction

Ever since the birth of comparative linguistics, officially though unjustly1 marked by Sir William
Jones’ proclamation of 1786 that Sanskrit Greek Latin and possibly Celtic Gothic and Old
Persian, as well, had “sprung from some common source perhaps no longer in existence”, the
primary goal of this scholarly discipline has been to uncover the prehistory and genetic relationship
of the languages studied. Doing so requires the application of, above all, the Comparative Method,
for a detailed description of which cf. section 2.1. What is generally looked for are linguistic
archaisms that can be projected back into the common and oftenmost no longer existing ancestor of
the related languages in question, i.e. their proto-language.
Following this fundamental philosophy of the scholarly field of comparative linguistics, the
present Ph.D. dissertation has sprung from a partly predefined subproject on archaisms in the
Germanic languages, i.e. those languages that can be scientifically proved by application of the
Comparative Method to derive from Proto-Germanic, at the University of Copenhagen based
research centre Roots of Europe – Language, Culture and Migrations which has as its proclaimed
goal to “create an open, lively and scientifically fruitful environment for the general study of
prehistoric Europe in a broad, interdisciplinary collaboration between scholars of comparative
linguistics, comparative religion, mythology and folklore, archaeology, genetics and
anthropology.”2
Such archaisms are expected to reveal the undisturbed development of linguistic segments from
a given proto-language to a subsequent linguistic stage of interest. In the case of Proto-Germanic,
linguistic archaisms are therefore expected to provide information on Proto-Indo-European, i.e. the
generally acknowledged parental language of Proto-Germanic, thereby also providing indispensable
information on Proto-Indo-European itself and its phonological and grammatical system, but the
benefits from knowledge obtained from linguistic archaisms are not limited to these areas. By

1 In 1767 around two decades prior to William Jones’ proclamation the Jesuit missionary Père Coeurdoux sent a
letter from his posting in Pondicherry to the French academy of science in which he outlined the similarity between
Sanskrit Greek and Latin and claimed them to have sprung from “une commune origine” since the five alternative
explanations of the resemblances between these three languages, viz. commerce, scholarly intercourse, proximity of
nations, religion and invasions, were to be disregarded, cf. e.g. Godfrey (1967: 58-59).
2 Cf. http://rootsofeurope.ku.dk/english/about_rootsofeurope/ (11 April 2014).

5
creating stronger links between, say, Proto-Germanic and its Proto-Indo-European precursor, the
scholarly community will also automatically obtain greater and more detailed knowledge on the
relations between Proto-Germanic, including the individual Germanic languages descended from it,
and the other Indo-European branches and languages of Europe and Western and Central Asia with
improved knowledge on prehistoric Europe in general as a secondary benefit.

1.1. Archaisms and innovations

In their fully understandable zeal to uncover archaisms in their respective languages of study,
comparative linguists often tend either to forget or at least to forget expressing overtly the necessity
of separating archaisms from innovations. In other words, in order to search for archaisms that can
be projected back into the proto-language, we must first define what constitutes an archaism. Only
so will we be able to identify and subsequently disregard innovations and their impact on the overall
language system in argumentations concerning the prehistory of the language or languages studied.
It should be noted that innovations are by no means useless to the field of comparative
linguistics. True enough, they may not provide the same kind of information on linguistic stages
past, but they constitute the primary means of determining linguistic subfamilies. Once it has been
established that two or more languages within a language family share an innovation, they can be
grouped together into a subfamily. In reality, however, subfamilies are not posited unless either
considerably many or very untrivial shared innovations can be identified between the languages
sharing them. If a language family is posited on the basis of only few, rather trivial shared
innovations, the risk must be considered that these innovations are tokens of mere chance
similarities rather than of a genuine linguistic subgrouping.

1.2. Aim and purpose of the present thesis

With the differences in focus and utility of linguistic archaisms and innovations borne in mind,
keeping these apart becomes all the more crucial to the field of comparative linguistics. Hardly
surprisingly, therefore, I have centred my Ph.D. project and the present thesis resulting from it on
the identification of archaisms and the subsequent separation of innovations from archaisms in
Germanic nominal morphology.
That I have chosen nominal morphology or more specifically derivational morphology of nouns
in Proto-Germanic and its descendants as my primary object of study in that regard is far from
desultory. Only few other fields of Indo-European linguistics are in equally great need for a study
on archaisms and innovations to be carried out. As I touch upon in greater detail in Ph.D. articles

6
no. 1 and 2, the inherited inflectional system of Germanic nouns collapsed and was shaped anew in
a manner better compatible with the Germanic phonological system that had also undergone
massive changes on its way from Proto-Indo-European to Proto-Germanic and further to the
individual Germanic languages. Going into details on the exact nature of and reasons for these
changes would take us too far; suffice it therefore here to state that the derivational stem of a given
noun, once easily recognisable by means of its mere inflection, had now become sufficiently
masked for the language users no longer to recognise this stem and consequently for the possibility
to arise that nouns could be assigned to a new inflectional class, a new pattern, thereby laying down
a seeming smokescreen on the derivational prehistory of these nouns, cf. e.g. Thöny (2013: 1, 23-
27, 41-43).
Two areas of the derivational morphology of Proto-Germanic nouns call for special mentioning
in this regard, viz. root nouns and vocalic stems. It is commonly known that a great amount of
fluctuation of nouns happened between the inflectional class of root nouns on the one hand and the
inflectional classes of the vocalic stems, i.e. the Germanic a- ō-, i- and u-stems, on the other as well
as internally between the vocalic stems, cf. e.g. Bammesberger (1990: 54-55), Casaretto (2004: 36,
170-173), Thöny (2013: 1-2, 13-27, 58-70, 77-82). Consequently, the mere assignment of a
Germanic noun to, say, the synchronic i-stem inflection does not vouch for its diachronic
identification as an i-stem, and we are forced to employ other means in order to uncover its
derivational history.

1.3. Structure and limitations of the present thesis

In two separate articles of my thesis, I have examined two inflectional and originally derivational
classes of Proto-Germanic nouns, viz. root nouns and primary i-stem nouns, in order to establish
new ways of defining which nouns of these two inflectional classes are archaic, i.e. which root
nouns and i-stem nouns have been inherited as such from Proto-Indo-European, and,
correspondingly, which nouns are to be regarded as innovations by way of having been secondarily
transferred into new inflectional classes differing from their original ones.
My preference for and choice of these two classes for examination relies on the circumstance
that, as will become evident to the readers from reading the articles, the phonotactics and ablaut of
the lexical roots from which their nouns are derived play a decisive role in the identification of
archaisms and innovations among their members. The same can rightfully be said of the primary
Germanic a- ō- and u-stems, but the comprehensive study on u-stems presented in the monograph
by Neri (2003: esp. 45-178, 341-346) has already, despite its primary focus on Gothic rather than on

7
Proto-Germanic as such, provided the scholarly community with invaluable information on
archaisms and innovations, and I see no reason to engage in a task already so well and thoroughly
conducted. As regards the two remaining classes of primary vocalic stems, viz. the a-stems and the
ō-stems, their very number far exceeds the spatial limitations of this thesis. Furthermore, the
complexity of the a-stems in particular as regards possible derivational sources even at the Proto-
Indo-European stage severely blurs our chances for finding any neat way to separate archaisms
from innovations in these two classes. A wide array of the derivational processes concerning
especially the Germanic a-stems has been dealt with in Ph.d.-article no. 2, though, and some of
those regarding ō-stems in Ph.d.-article no. 1.
Diachronic derivational morphology as a field of study cannot be properly conducted without
accurate inputs from diachronic phonology. In other words, if we do not know the basic recurrent
phonological correspondences between the languages compared, we have no means of determining
whether two forms from two different languages are cognate or not. Consequently, in order for us to
better understand all derivational and inflectional aspects of the Proto-Germanic root nouns and
primary i-stem nouns, I have deemed necessary the addition of two auxiliary articles on
phonological questions to my thesis: one on the Germanic outcome of initial PIE *#Hi- and *#Hu-,
the other on the Germanic outcome of PIE *- and - in initial syllables. Besides
suggesting a new soundlaw regarding the development of Proto-Indo-European laryngeals into
Proto-Germanic, the former article is meant to qualify the Indo-European etymology of one of the
lexemes included in Ph.D. article no. 4 on root nouns, viz. PG *aik- f. ‘oak’.
As for the latter article, the sound laws suggested in it have a direct relevance to our
understanding of the prehistory of some of the oblique desinences of the Proto-Indo-European i-
and u-stems, thereby contributing to our overall understanding of these two inflectional classes and
the potential impact of their inflectional features on common points of reference to other
inflectional classes.

1.4. General features of the individual articles

As these four articles are aimed at publication in various media, they cannot be expected by their
readers to appear entirely uniform. Nevertheless, it is worth mentioning several recurrent traits of
especially those three articles where etymological analyses play a significant role.
The etymological analyses of individual words are always initiated with the specification of the
reconstructed or, in some cases, transposed Proto-Germanic lemma in question. The lemmata are
quoted in their stem form, i.e. in a form clear of further inflectional specifications, meaning that,

8
e.g., the noun known as PG *dagaz (nom.sg.), *dagan (acc.sg.) etc. ‘day’ will be quoted in the form
of *daga-. Following the notational system introduced by the more recent etymological dictionaries
of Proto-Germanic, e.g. Orel (2003) and (Kroonen (2013), I mention verbs by their infinitive stem.3
Consequently, the verb whose infinitive is PG *kwemanan ‘(to) come’ will be quoted as *kwemana-
; not as *kwema- in accordance with the notational practice followed by e.g. Seebold (1970) and
Casaretto (2004). Immediately following the Proto-Germanic lemma itself, basic grammatical
information is provided, i.e. information on word class. Nouns are marked and defined by gender;
verbs according to their type, i.e. whether they belong to the class of strong, weak or preterite-
present verbs. In cases where, for some reason or another, the grammatical gender of a noun cannot
be identified the noun is marked as “sb.” for substantive “n.” being reserved for marking the
neuter. The final piece of information provided on every lemma is its meaning. If, as is the case for
some glosses, the meaning is only provided in, say, Latin or Greek, I have generally attempted at
providing also an English translation.
Attested cognates of the Proto-Germanic lemmatа in the individual Germanic languages,4 further
Germanic derivatives related to the Proto-Germanic lemmata, and considerations on the Proto-Indo-
European predecessors of the lemmata constitute, in the sequential order mentioned here, the next
pieces of information provided in the etymological analyses. All intra- and extra-Germanic cognates
are generally listed in the same way as the Proto-Germanic lemmata, i.e. with their form followed
by basic grammatical and semantic information. Their forms are quoted in accordance with the
standard annotational practices of the languages in question, e.g. Sanskrit nouns and adjectives in
their stem form, Sanskrit verbs in their 3.sg., Greek nouns in their nom.sg., Latin adjectives in their
m.nom.sg., Gothic verbs in their infinitives etc. Generally, the cognates mentioned in my thesis
belong to the oldest attested stages of the languages which they represent unless later stages provide
indispensable pieces of etymological information not provided by the ancient languages.
The etymological analyses are concluded by a list of references to secondary literature regarding
the Proto-Germanic lemma in question. Reference is given to a range of standard handbooks on

3 Orel (2003) quotes the full infinitive form, i.e. stem and desinence. In the case presented here, Orel (2003: 227)
quotes the form as PG *kwemanan.
4 I generally list cognates from only the major ancient Germanic languages, i.e. Gothic, Old Norse, Old English, Old
Frisian, Old Saxon and Old High German, unless features of cognates from later stages of these languages or from
other of the ancient Germanic languages provides unique information of particular relevance to the etymological
analysis of the lexeme in question.

9
Germanic and Proto-Indo-European etymology, viz. IEW (2005), LIV (2001) and NIL (2008) for
Proto-Indo-European; Orel (2003), Kroonen (2013), Seebold (1970), Bammesberger (1990),
Heidermanns (1993), Schaffner (2001) and Griepentrog (1995) for Proto-Germanic; Lehmann
(1986) and Casaretto (2004) for Gothic; De Vries (1962) for Old Norse; Holthausen (1974) for Old
English; Boutkan & Siebinga (2005) for Old Frisian; Sehrt (1966) for Old Saxon; Lloyd & Springer
(1988), Lloyd et al. (1998) and Lloyd & Lühr (2007) for Old High German; and Bjorvand &
Lindeman (2000), Philippa et al. (2003-2009) and Kluge/Seebold (2002) as representatives of
modern Germanic language etymological dictionaries with independent and original etymological
proposals. These dictionaries also serve as my main source for data.
As for the wide array of additional handbooks also consulted, e.g. synchronic dictionaries on the
individual Germanic languages and synchronic or etymological dictionaries on languages of the
remaining Indo-European branches, I have generally mentioned them in the list of references only
when parts of the etymological analyses of the Proto-Germanic lemmata in question rely on
etymological proposals mentioned in them.
Finally, it should be noted that I have merged the bibliographical sections of the four articles and
of the general chapters 1-3 into one list situated at the end of the thesis.

10
2. Methodological considerations

Before we turn to the four individual articles, we need first identify the methods applied to and the
factors operating in their data. The most salient such factors are regular sound change as identified
by the concurrent application of the Comparative Method and the method of Internal
Reconstruction, analogy, and lexical borrowings.

2.1. The Comparative Method

The by far most important method in historical and comparative linguistics and consequently also
for the present thesis is the Comparative Method. When applied on linguistic data from two or more
languages of common descent, i.e. two or more languages sharing the same ancestor, the
Comparative Method makes us capable of revealing the development of these languages by
performing a segment-by-segment comparison of the data. For a thorough description of the
Comparative Method cf. e.g. Campbell (2013: 107-144).
The linguistic area of phonology constitutes the primary source of data for comparison.
Consequently, our task as comparative linguists is to compare lists of cognate terms from which we
can then establish regular sound correspondences between the languages in question. Only when a
sequence of regular and recurring sound changes can be suggested can we reconstruct the proto-
language, i.e. the common ancestor of the languages studied. In other words, linguistic kinship is
deemed certain only if the common ancestor appears reconstructible and if regular and recurring
sound correspondences can be established. After establishing these and reconstructing proto-
phonemes, we can turn our focus one level up in the linguistic hierarchy and start comparing
meaning-bearing sequences of these phonemes or, in other words, start comparing morphemes. In
so doing we must be constantly aware, though, that the results of morphological comparison are
often seriously hampered by analogy, for an outline of which cf. section 2.3.

2.1.1. How to apply the Comparative Method on linguistic data

Campbell (2013: 109-128) suggests some basic steps for applying the comparative method to
linguistic data, the first of which being the creation of a list of potential cognates, i.e. a list of
linguistic forms with identical or similar meaning from the languages in question.
Secondly, we must establish sets of regular and recurrent phonological correspondences. If we
can retrieve many regular correspondence sets of this kind, a common origin of the languages

11
involved in the comparison becomes a virtual certainty, especially if some of the correspondences
are non-trivial or unusual.
The reconstruction of proto-phonemes constitutes the third task. One type of consideration that
might assist us in deciding what reconstruction fits the data is of a typological nature. For example,
the voicing of voiceless stops between vowels is common, but not the devoicing of voiced stops in
the same position. If, say, a correspondence -p- ~ -b- is found intervocalically in two languages, the
proto-phoneme is more likely to be *-p-, with the second language having undergone the
development *-p- > -b-. The opposite reconstruction would create a rare type.5
Fourthly, we might need applying contextual conditionings to the sound correspondences just
discovered, i.e. we must state if two or more sounds appear to be complementarily distributed
according to the surroundings in which they occur. In other words, if, say, a k of one language can
be compared to both a k and a of another language, we must either assume that original *k and *
have merged in the former or that original *k (or * for that matter) has split into what ultimately
developed into two phonemes in the latter. In this latter case, we need to set up a range of criteria
for when to expect k and when to expect , i.e. we need to identify the conditioning factor.
Steps five and six partly overlap and strongly relate to parts of the process in step three. Here we
must check the reconstructed proto-phonemes for typological adequacy from the perspectives of
both the reconstructed phonological system of the proto-language on the one hand and linguistic
universals and typological expectation on the other.
When we have accomplished all the previous steps, we can safely turn to the comparison of
larger units such as morphemes, cf. above.

2.1.2. Potential shortcomings of the Comparative Method

Criticism of the Comparative Method has mainly centred on the, admittedly valid, observation that
languages do not develop as divergently as inherently presupposed by the Comparative Method or
rather as illustrated by the family-tree model of rendering linguistic kinship as a branching off in
different directions of the related languages from one common source, cf. e.g. Campbell (2013:
142-144, 174-184, 187-188) and the scepticism expressed by Bloomfield (1933: 318): “The
comparative method, then – our only method for the reconstruction of prehistorical language, –

5 Unusual sound changes do occur, though. The Indo-European word for e.g. ‘two’ is reconstructed as PIE *d
which, in turn, is reflected as Arm. erku num. ‘two’. Several other cognates demonstrate a regular change of PIE
*d - > Arm. erk-, cf. e.g. Fortson (2004: 341-342).

12
would work accurately for absolutely uniform speech-communities and sudden, sharp cleavages.
Since these presuppositions are never fully realized, the comparative method cannot claim to picture
the historical process.”
In reality, however, linguists are well aware of this caveat and therefore, when possible, seek to
include data from variants of the languages studied, cf. e.g. Campbell (2013: 188-197). Such data is
not easily rendered in the family-tree model which, in itself, allows for the rendering of neither
contact between languages after their first split-up nor the circumstance that no languages are clear
of dialectal or other types of linguistic variation. In order to render such data properly in a graphic
model, we must apply the Wave Theory according to which every linguistic change spreads out like
a wave from its starting point. Consequently, we must insert every change and its geographical area
of impact into a map, thereby obtaining a detailed picture of linguistic variation and contact.
It should be noted, however, that the Wave Theory reveals nothing about linguistic stages past.
The most comprehensive and accurate rendering of the history of a given language, consisting as it
does of variants and influenced as it probably is by neighbouring speech communities, is therefore
obtained by application of both models whenever possible.

2.2. Internal Reconstruction

Whereas the Comparative Method performs a segment-by-segment comparison of linguistic data
from two or more genetically related languages, the method of Internal Reconstruction analyses the
internal development of only one language over time. In practice, these two methods should by no
means be seen as mutually exclusive. On the contrary, historical and comparative linguists apply
both methods together in order to reconstruct prehistoric phases of languages, to discover the
linguistic developments of the languages studied and to confirm or refute hypothesised kinship
between languages, cf. e.g. Campbell (2013: 211).
Internal Reconstruction is a method of retrieving information about a past and often most
unattested stage of a language from the characteristics of an attested and more recent stage of that
language. Internal reconstruction thus compares variants within a single language under the
assumption that meaning-bearing elements alternating between two or more similar forms in
different environments, i.e. different allomorphs of a morpheme, descend from a single regular form
from which they have arisen as the result of conditioned sound changes, cf. e.g. Campbell (2013:
198-212) and Givón (2000: 107-114, 155).

13
2.3. Analogy
Analogy is best described as a cognitive process of transferring or generalising information or
meaning from one particular set of conditions to another particular set of conditions, cf. e.g.
Campbell (2013: 91-105) who lists a range of different types of analogy including, above all,
proportional analogy and analogical or paradigmatic levelling.
Such a definition of analogy implies that analogy results from a cognitive and in principle
irregular and unpredictable process rather than from an automatic or regular one as is the case with
the regular and recurring sound changes identified by means of the Comparative Method and
Internal Reconstruction. The relationship between analogy and sound change is described quite
accurately by what is known as Sturtevant’s (1947: 109) Paradox: “Phonetic laws are regular but
produce irregularity. Analogic creation is irregular but produces regularity.” Hence follows that
analogy actually modifies the outcome of regular sound change, thereby disturbing or blurring the
clarity and regularity of the lists of cognates used by the Comparative Method to identify regular
sound changes and to establish linguistic kinship.

2.3.1. Proportional analogy

One type of analogy is the proportional one which is the linguistic process that removes word forms
perceived by the language users as irregular by remaking them in the shape of more common forms
governed by synchronic rules, cf. e.g. Campbell (2013: 92-93).
The idea is that an old synchronically irregular form can be replaced by a new synchronically
regular and transparent one by application of the formula a : b ~ c : x where x is to be understood as
a new form whose phonological shape resembles that of b. Suffice it here to mention one example,
viz. that of turning a Germanic strong verb perceived as irregular by the language users into a
synchronically more regular weak verb, e.g. Da. male ‘paint’ : gale ‘crow’ ~ malede ‘painted’ : x
where x = galede ‘crew crowed’ the obsolete form being gol ‘id.’. Sometimes irregular forms can
be created by analogy as well, this process being referred to by Campbell (2013: 95) as analogical
extension. One case is the past tense form of Eng.(Am.) dive, i.e. drive : dive ~ drove : x where x =
dove, the obsolete form being dived still existing in British English.

2.3.2. Analogical or paradigmatic levelling

Another form of analogy is analogical or paradigmatic levelling by which process the language
users make a synchronically irregular paradigm with variants regular by choosing and generalising
one of the variants, cf. e.g. Campbell (2013: 93-95). Consequently, we may justly state that both

14
cases of proportional analogy described above (section 2.3.1) are also cases of analogical or
paradigmatic levelling.
One of the cases of analogical or paradigmatic levelling most often adduced in handbooks on
linguistic change is that of Gr. ἕπομαι ‘follow’ whose present singular and plural forms if resulting
from regular sound change only, should have been 1.sg. ἕπομαι ‘I follow’ 2.sg. *ἕτηι ’you follow’
3.sg. *ἕτεται ‘he follows’ 1.pl. ἑπόμεθα ‘we follow’ 2.pl. *ἕτεσθε ’you follow’ 3.pl. ἕπονται ’they
follow’. In order to create a uniform paradigm the stem variant *ἕτ-, resulting from the
development of PIE *kw > Gr. τ when followed by a front vowel, was replaced with ἑπ- whose π
reflects PIE *kw in front of back vowels, cf. e.g. Campbell (2013: 94) quoting Beekes (1995: 73).

2.3.3. Analogical processes of particular relevance to the present thesis

As I stated above (section 1.2.), my thesis centres on the identification of archaisms and the
subsequent separation of innovations from archaisms in Germanic. Analogical processes are,
therefore, of the utmost importance to its argumentation and results.
Analogy is responsible for the creation of innovations. In other words, when an archaic- and
irregular-looking Proto-Germanic paradigm is analogically altered into a more regular one in
synchronic terms, we are facing a linguistic innovation. Archaisms, on the other hand, must be
regarded as resulting from processes of regular sound change that have been allowed to remain
undisturbed by analogy.
Derivational morphology of Germanic nouns is a field of study for which analogy and the
linguistic innovations resulting from it play a significant role. Due to the collapse of the inherited
inflectional system of Germanic nouns, i.e. a system in which lexical root, derivational suffix and
inflectional ending were originally more easily (though not entirely) recognisable units, the
Germanic nouns have lost their primary marker of derivational stem and inflectional class, these
two entities being originally identical. In other words, when, say, the easily recognisable a-stem PG
*dagaz m. ‘day’ is reflected in the individual Germanic languages as Goth. dags m. ‘id.’ ON dagr
m. ‘id.’ OE dæg m. ‘id.’ etc. with regular syncope of the stem defining vowel PG *a, we are
suddenly deprived of our most straightforward means of categorising these nouns diachronically,
i.e. of identifying them as a-stem derivatives, cf. e.g. Thöny (2013: 41-43, 53-58) and Nübling
(2008: 283-288, 295-297) for a more comprehensive discussion of the impact of the loss.6 7 If, then,

6 Though including a section of approx. two pages on Proto-Germanic inflectional classes, Nübling (2008) treats the
matter of inflectional classes primarily from the view of contemporary German.

15
we add to the equation the i-stem PG *gastiz m. ‘guest’ > Goth. gasts m. ‘id.’ ON gestr m. ‘id.’
OE giest m. ‘id.’ etc. with regular syncope of the stem defining vowel PG *i, a classical case of
inflectional similarities between two inflectional classes emerges. Especially in Gothic where the i-
mutation does not operate, the so-called pivotal forms of nom.sg. Goth. dags : gasts and acc.sg. dag
: gast now form the basis for a merger of the remaining desinences, in this case in favour of the a-
stem forms. This and similar examples constitute what Ramat (1981: 62) designates a “Krise des
Flexionssystem im Germanischen.”
However, crises are meant to be solved. The Germanic languages found new ways of deciding to
what inflectional class a given noun should most probably be assigned. Though often deprived of
their overt derivational markers, the members of each inflectional class still inherited a wide array
of features from older linguistic stages, and these features assisted in the creation of more or less
fixed prototypes or profiles of inflectional classes, cf. e.g. Thöny (2013: 23-27, 35-46, 314-325, esp.
317) who lists features such as sets of inflectional rules (e.g. desinences and mutations of the stem),
phonotaxis (number and quantity of syllables; to this I add quality of syllables, i.e. ablaut), stress
and grammatical gender among the formal ones. Among the functional features, he lists general
function (agent noun, abstract noun etc.), semantic field (+/- animate, +/- animal etc.), natural
gender and semantic relations to the derivational basis.8
In cases where doubt prevails as to the proper assignment of a given noun to an inflectional
class, the loss or weakening of the once primary marker of derivational stem as well as inflectional

7 Following Darski (2004), Kotin (2012: 93-101, 451-452) discards the entire standard notational system of a-stems, i-
stems, u-stems etc. and replaces that system with one of referential nominal stems (RNS) consisting of referential
nominal roots (RNW) and primary, constant exponents (KE) to which secondary, variable exponents (VE) are
added. In this system, the RNW broadly corresponds to the lexical root, the KE more or less to the derivational
suffix and the VE to the inflectional ending. Thus, Kotin (2012: 101) designates a classical a-stem as “RNS 1: RNW
+ KE idg. O, germ. A” with the result of focus shifting from (diachronic) derivation to (synchronic) inflection in a
way quite similar to the inflectional classes of Thöny (2013: 15-20) and others, cf. Hansen (forthc. 2015: 123).
Kotin’s practice can therefore hardly be seen as anything but an attempt to adapt the notational system to the
synchronic reality of the ancient Germanic languages.
8 Albeit only to a limited extent when compared to the Proto-Germanic situation, the individual Germanic languages
still include the primary marker of a given inflectional class among the formal features for assignment of nouns into
inflectional classes. The relatively stable Old English r-stems constitute a brilliant example of this. With their rather
consistent suffixal r, they defy the transitional tendencies seen in many other inflectional classes for a considerable
period, cf. e.g. Thöny (2013: 43, 84-85).

16
class, i.e. the loss or weakening of the derivational stem marker, makes the remaining features gain
attractive power as regards this assignment.

2.4. Lexical borrowings

At least one additional factor potentially disturbing the picture obtained by the application of the
Comparative Method deserves to be mentioned here, viz. language contact including lexical
borrowings which, naturally, cannot be reconstructed any longer back in time than to the period
when they were borrowed from the source language into the borrowing language, cf. e.g. Campbell
(2013: 56-75).

2.4.1. Lexical borrowings from known sources

In many cases, the identification of lexical borrowings involves no particular challenges to

comparative linguistics. When, for instance, OE cyrice, cirice f. ‘church’ and OS kirika f. ‘id.’ has
been borrowed from Gr. κυριακόν, κυρικόν n. ‘pertaining to the Lord’ we can easily identify the
Germanic terms as borrowed rather than inherited term. The reasons for this are manifold. Firstly,
we find no suitable extra-Germanic comparanda displaying a regular phonological development; in
other words we find no reflexes of a PIE †gurigeh2 in any Indo-European languages but Germanic
(i.e. PG †k r k n -). Secondly, a nearly identical term is found in Greek where, unlike in
Germanic, a perfect Indo-European etymology can be adduced.9 Thirdly, early Germanic
attestations are found in only a limited area of Germania. The word for ‘church’ has thus been
borrowed from the West Germanic into the North Germanic languages as is evident from the fact
that a form such as PG †k r k n - would probably result in ON †kurka rather than kyrkja, kirkja
n
just as PG *tal d w.v. I ‘told; counted’ (pret.) > ON talda and not †telda, cf. e.g. Noreen (1923:
61-63). This explanation conforms well to what we know about the culture of the early Germanic
area, viz. that the speakers of North Germanic were Christianised considerably later than the
speakers of, say, Old English, cf. e.g. Stenton (1971: 104-105) and McGuire (2008).

2.4.2. Lexical borrowings from unknown sources

However, not every case of lexical borrowing is as easily identifiable as the above-mentioned
(section 2.4.1.). If we suspect a given lexeme to be a lexical borrowing in cases where the source

9 Gr. κυρι(α)κόν is to be analysed as a substantivised ko-adjective derived from Gr. κύριος m. ‘possessor master lord’
< PIE *k̂ H-r o- to the root PIE *k̂e H- ‘swell’ also found in e.g. Gr. κυέω v. ‘swell’ and Skt. śávīra- adj. ‘strong
mighty’ cf. e.g. Nielsen (2000: 220-221), Frisk (1963-1966: 53-54) and IEW (2005: 592-594).

17
language is no longer attested, we are left with no other choices than resorting to indications such as
limited geographical distribution of the lexeme in question, phonological and morphological
irregularities, aberrant processes of word formation and the connection of the lexeme to a semantic
field especially prone to accept lexical borrowings, cf. e.g. Schrijver (1997: 294-296) and Campbell
(2013: 61-66). As pointed out by Schrijver, the identification of a given lexeme as a borrowing may
justly call for criticism if lexical borrowing from an unknown source is invoked only by means of
one of the criteria listed here. Rather as Schrijver (1997: 296) states “[i]t is the cumulative
evidence that […] clearly tips the balance in favour of a substratum origin.” This is especially true if
we add that cases of lexical borrowing from unknown sources can be identified as such “by the fact
that they show phonological and morphological alternations which are regular in the sense that they
recur in more than one etymon according to a certain pattern but irregular in the sense that they
cannot be explained, for some reason or other, on the basis of Indo-European phonology and
morphophonology” cf. again Schrijver (1997: 296).
Lexical borrowings from unknown sources are generally referred to as sub-, ad- or superstrate
words. The term substrate or substratum refers to the language of speakers who had less prestige
and power in comparison with speakers of another language in contact with the former. Conversely,
the term superstrate or superstratum is used of a language whose speakers enjoy high prestige in
comparison with speakers of another language upon which the former language has often been
imposed by means of force, i.e. invasion. Adstrate or adstratum finally labels the linguistic
interaction of equally prestigious speakers of two or more different languages, cf. e.g. Tristram
(2007: 195-196).
When we are in want of historical sources to inform us about the balance of power and prestige
between the languages in contact, our only tool for categorising a lexical borrowing as either a sub-,
ad- or superstrate word is an analysis of the semantic field to which it belongs. Superstratum is
invoked when the lexeme in question belongs to a semantic field pertaining to upper class functions
of a society such as religion, administration and warfare; substratum, on the other hand, when the
lexeme belongs to a semantic field pertaining to lower class life such as farming and nature in
general. To these semantic fields of substrate vocabulary, Vennemann (1995 [2003]: 207-208) adds
that of toponyms assigned to various locations in the area where the speakers of a superstrate
language impose themselves on the speakers of a substrate language. Furthermore, he notes that
substrate languages tend to influence not only on the lexicon but also on the morphosyntax and
phonology of intruding superstrate languages.

18
2.4.3. Borrowing processes of particular relevance to the present thesis

Since lexical borrowing constitutes a natural part of any language that finds itself in contact with
other languages, cf. e.g. Campbell (2013: 56), I see no reason not to assume that Proto-Germanic
and the ancient Germanic languages could have borrowed lexical items from other known or
unknown languages.
Schrijver’s (1997: 296) statement that lexical borrowings can safely be identified in cases where
recurring phonological and morphological patterns unanalysable within the phonological and
grammatical system of the target language are displayed has provided us with a new and viable tool
for suggesting substrate or loan word origin of Germanic lexemes. Possible candidates for lexical
borrowing according to these methodologically strict criteria have been suggested by Schrijver
(1997: 297-312) himself and repeated later by e.g. Kroonen (2012: 240). He has observed a
systematic, though non-Indo-European interchange of initial *a- versus initial *Ø- in a number of
lexically similar doublets from the European languages, cf. e.g. OHG amsala f. ‘blackbird’ ~ Lat.
merula f. ‘id.’(< *mesal-) and OHG aruz n. ‘ore’ (< PG *arut-) ~ Lat. raudus m. ‘lump of ore’. We
can further observe that the lexical roots display radical zero grade when a-prefigated.
Another case identified already by Kuiper (1956: 217-219) is that of the suffix PG *-īt- (<
Pre-PG *-ind-) in e.g. PG *arwīt- f. ‘pea’ corresponding almost regularly to the “Pelasgian” suffix
Gr. -ινθ- ~ - - ~ - θ- ~ -ιν- etc. found in e.g. Gr. ἐρέβινθος m. ‘chickpea’ λαβύρινθος m. ‘labyrinth’
etc. This suffix seems to occur also with a rounded vowel, i.e. as Gr. -υνθ- found in e.g. Gr.
κολοκύνθη f. ‘round gourd’ ὀ όλυνθοι m. ‘chickpeas’ (pl.) etc. cf. also Kroonen (2012: 243-244,
247-248).

19
Article no. 1: Layers of root nouns in Germanic:
Chronology, structure and origin

In this article,10 I propose that root nouns and unsegmentable consonant stems in
Germanic be categorised into three layers, viz. (I) inherited root nouns, (IIa) original
substrate or loan words, (IIb) nouns from other declensions reanalysed as root nouns
in parts of the Germanic dialect continuum and (III) nouns from other declensions
reanalysed as root nouns in North Germanic. I further propose that the inherited root
nouns of the first layer, non-ablauting as they have become in Germanic, display
radical ablaut grades that are predictable from the phonotactic structure of the root.

1. Communis opinio on root nouns

Proto-Indo-European root nouns are constructed only by means of a root and an inflectional
morpheme. Thus, no derivational suffixes are involved in the construction of a root noun. Like the
majority of the remaining athematic noun declensions, root nouns are ablauting, i.e. their radical
vocalism interchanges between -grade ō-grade and zero grade according to a number of
morphophonemic criteria from Pre-Proto-Indo-European times, cf. e.g. Brugmann & Delbrück
(1906: 130-146), Hirt (1929: 224-230), Kuryłowicz (1968: 26-38), Schindler (1972a: 8-9);
Rasmussen (2003: 354-355); Fortson (2004: 73-74, 103-109) etc.
According to Schindler (1972: 32-38), root nouns display two basic types of ablaut at the Proto-
Indo-European stage.

1 e/Ø-ablaut (e.g. PIE nom.sg. *h2n r, acc.sg. h2n r- , gen.sg. *h2nr-ós ‘man’). This type is
associated with action nouns and with agent nouns derived from verbal roots with inherent
stative semantics.
2 o/e-ablaut (e.g. PIE nom.sg. *nókwt-s, acc.sg. *nókwt- , gen.sg. *nékwt-s ‘night’); however
o/Ø-ablaut if the root contains a vocalisable resonant (e.g. PIE nom.sg. * rk̂-s, gen.sg.
* rk̂-ós ‘fallow deer’). This type is associated with feminine nouns with resultative or

10 An abbreviated version of this article has been submitted for publication in the conference proceedings from
Etymology and the European Lexicon: XIII Fachtagung der Indogermanischen Gesellschaft, Copenhagen, 17-22
September 2012.

20
passive semantics and with agent nouns with which a touch of iterative semantics can often
be assocoiated.

Root nouns and consonantal stems in general were inherited from Proto-Indo-European into Proto-
Germanic, where, however, this declension had become not only closed in that it was no longer
productive and would not accept new members, but also moribund in that its members, the inherited
root nouns, gradually transcended to other declensions.
This communis opinio is expressed by, i.a., Krahe (1967: 34): “Von dem im Germ. vorhandenen
kons. Stämmen die sämtlich auf idg. Typen beruhen stellen die unter ”a” bis ”d” [Reste von idg.
Wurzelnomina u. dgl., Verwandtschaftsnamen auf -r, Substantivierte Partizipia auf idg. -nt and
Reste neutraler s-Stämme] behandelten Restgruppen dar, die im Laufe der Entwicklung in den
Einzeldialekten als selbständige Gruppen ausgestorben und in andere Klassen übergegangen sind.”
Two general circumstances call for a modification of this view. First, a fact long known but too
often left unmentioned by the scholarly community is that the root noun declension seems to have
become (re)opened in North Germanic, cf. e.g. Brøndum-Nielsen (1935: 146): “De maskuline
Rodstammer udgør i Nordisk en faatallig Klasse, af hvilke næppe andre end fótr Fod er oprindelig
Rodstamme.” and (1935: 154-155): “De feminine Rodstammer omfatter i Gammeldansk – som i de
andre nordiske Sprog, dels en Række gamle Rodstammer, dels nogle fra andre Stammer overførte
Substantiver.”
Second, Kroonen (2012: 255) claims that: “[…] in Germanic the Indo-European class of the root
nouns was open to loan words or substrate words. This can hardly be anything else but a reflection
of the form of these words in the donor language.”
If so far taking for granted the validity of these two assumptions, we seem able to conclude that
the root nouns had become if not thriving in Germanic then at least in the process of recovering
from their former moribundity. Consequently, a rejection or at least a modification of the communis
opinio regarding the fate and the vitality of the root noun declension seems to be in order.
In this article, I claim the alternative view that root nouns and synchronically unsegmentable
consonant stems be stratified into a number of layers according to various criteria. Basing my
analysis on the Germanic material listed in section 2, I propose the following three layers:

I Root nouns inherited from Proto-Indo-European.

IIa Unsegmentable substrate or loan words reanalysed as root nouns in Proto-Germanic.

21
IIb Nouns from other declensions reanalysed as root nouns in Proto-Germanic or in parts of
the Germanic dialect continuum.
III Nouns from other declensions reanalysed as root nouns in North Germanic.

2. Material
The hopefully close to complete list of material below consists of 69 lexemes that are inflected
entirely or partially as root nouns or as unanalysable, monosyllabic consonant stems in Proto-Indo-
European (only those that are continued in Proto-Germanic), Proto-Germanic or the individual
Germanic languages.

1 PG *aik- f. ‘oak’. Attested as a root noun in ON eik f. ‘id.’ OE c f. ‘id.’; as an ō- or i-

stem in OFris. k f. ‘id.’ OS k f. ‘id.’ (may also be a root noun) as an i-stem in OHG
ei(c)h f. ‘id.’ (may also be a root noun) and as a younger ō- or iō-stem in OHG eihhe, eihha
f. ‘id.’. Often affiliated with the root PIE *h2e ĝ- ‘shine’ cf. Gr. αἰγι- (e.g. in αἰγίλωψ m.
‘kind of oak’) Gr. αἴγειρος m. ‘poplar’ Lat. aesculus f. ‘durmast oak winter oak’ vel sim.
(< *aigscolos), Lith. ąž olas, á ž ols, á ž olas m. ‘oak’ etc.; further maybe ORu. jazvъ m.
‘badger’ Ru. jazь m. ‘carp’ OIr. áesc sb. ‘concha clasendix’. Rather than reconstructing
PIE *h2e ĝ-, we might consider reconstructing *a ĝ- with an original (post-) PIE *a, this
alternative being more likely when judging from the semantics (botany) which at least
partially implies a possible loan word status and thus affiliation with layer IIa rather than
layer I, cf. also e.g. Kroonen (2013: 9-10) for the Germanic forms and Frisk (1960: 30-31)
for the Greek cognates. Literature: Bammesberger (1990: 195), Bjorvand & Lindeman
(2000: 180), Griepentrog (1995: 24-32), Holthausen (1974: 2), IEW (2005: 13),
Kluge/Seebold (2002: 229-230), Kroonen (2013: 9-10), Lloyd et al. (1998: 974, 984-986),
Orel (2003: 7), Philippa et al. (2003: 669), de Vries (1974: 96).
2 PG *alh- m. ‘temple sanctuary’. Attested as a root noun in Goth. alhs; as an a-stem in OE
ealh, alh m. ‘temple place of sacrifice protected area’ OS alah m. ‘Christian temple’;
further cf. the compounded forms ON -áll and OHG alah-. Possibly to be reconstructed as
PIE *h2élk- ~ *h2lk- in light of Gr. ἀλκί m. ‘strength’ (dat.sg.); semantic connection to PG
*alh- rather weak, though. An o-stem PIE *h2ólk-o- also underlying the Germanic a-stem
is attested in Celt. alko-, Lith. alkas, elkas m. ‘sacred grove’ and Latv. lks m. ‘idol’, and
an i-stem is found in compounded forms such as Gr. ἀλκί- and Celt. alki-. Further

22
connections to the root PIE *h2lek̂s- ‘ward off guard protect’ have also been invoked cf.
e.g. Skt. rák at v. ‘protects’ and Gr. ἀλέξω v. ‘ward off’. A possible semantic point of
reference between the Germanic and Balto-Slavic forms (‘sacred grove temple’ etc.) on
the one hand and Gr. ἀλκί m. ‘strength’ as well as PIE *h2lek̂s- ‘ward off guard protect’
on the other hand may be seen in OE ealh, alh ‘temple, place of sacrifice protected area’.
Kroonen (2013: 22) prefers to regard the Germanic and Balto-Slavic forms as borrowed
from a local non-Indo-European language with the consequence that PG *alh- would
belong to layer IIa rather than I or IIb. Literature: Bammesberger (1990: 195), Casaretto
(2004: 38-39), Griepentrog (1995: 33-57), Holthausen (1974: 84), IEW (2005: 32),
Kroonen (2013: 22), Lehmann (1986: 27), LIV (2001: 278), Lloyd & Springer (1988: 138-
139), Orel (2003: 14), Sehrt (1966: 9), De Vries (1962: 6-7).
3 PG *and-, *anþ-, *und-, *unþ-, *umbi prep. ‘about’ etc. with a wide array of descendants
in the individual Germanic languages. To be analysed as fossilised case forms of PIE
*h2ént- ~ *h2nt- ‘front forehead’ cf. Hitt. ha-an-za /hant-s/ sb. ‘front’ etc., with the
original ablaut apparently preserved in Germanic. Outside Anatolian, this root noun is used
adverbially in fossilised case forms. Literature: Griepentrog (1995: 443-444), IEW (2005:
48-50), Kroonen (2013: 31, 558-559), Orel (2003: 18).
4 PG *anad-, *anid-, *anud- f. ‘duck’. Attested as a root noun only in ON ǫnd, ǫnð f. ‘id.’;
as an i-stem in OE ænid, æned, ened m./f. ‘id.’, OS anad, anud f. ‘id.’ and OHG anut, anat
etc. f. ‘id.’. To be reconstructed as PIE *h2enHt -i)- ~ *h2nHt -i)- ‘duck web-footed bird’
cf. also e.g. Skt. tí- f. ‘duck web-footed bird, aquatic bird’ Gr.(Ion.) νῆσσα f. ‘duck’,
Gr.(Att.) νῆττα f. ‘id.’ Lat. anas f. ‘id.’ (gen.sg. anatis), Lith. ántis f. ‘id.’, OPr. antis f.
‘id.’. The comparative data suggest this lexeme to be either an i-stem or a root noun. In the
former case, ON ǫnd, ǫnð must be analysed as secondarily transferred to the root noun
declension and thus belonging to layer III. The attestation of a root noun in Latin, however,
severely weakens any argument in favour of PG *anad-, *anid-, *anud- being an original
i-stem. In the latter case, i.e. if it is to be analysed as an original root noun belonging to
layer I, at least a partial transition from root noun to i-stem inflection seems to have taken
place in Proto-Indo-European already, for which process cf. also Ph.D. article no. 2. The
presence of the enigmatic second vowel PG *a ~ *i ~ *u may be regarded as a further
support for a root noun origin of the lexeme in question. Hamp (1978: 30) thus believes to
see this second vowel of the lexical stem as a “continuation of non-initial non-medial

23
schwa”.11 In contrast to e.g. Griepentrog (1995: 299-300) and Hollifield (1984: 34-36),
Fulk (1988: 153-154) and tentatively Hansen (2007: 157-158) support Hamp’s analysis
but the former adds that the development cannot be restricted to final syllables in light of
PG *anad-, *anid-, *anud- being, in his view, an i-stem. Alternatively, it can be analysed
as an originally ablauting t-stem PIE *h2énh2-et- ~ *h2nh2-t-, cf. Kroonen (2013: 26) and
Beekes (1985: 63-64) for the assumption of two generalised full grades. Belonging to layer
I or, in the light of its appearance as a root noun only in Old Norse, layer III. Literature:
Bammesberger (1990: 202-203), Bjorvand & Lindeman (2000: 34-35), Holthausen (1974:
91), IEW (2005: 41-42), Kluge/Seebold (2002: 246), Kroonen (2013: 26), Lloyd &
Springer (1988: 291-293), Orel (2003: 21), Philippa et al. (2003: 659), De Vries (1962:
687).
5 PG *b k- f. ‘beech; book’. Attested as a root noun in ON bók f. ‘id.’, OE b c f./n. ‘book’
and OS b k f./n. ‘id.’, OHG buoh m./f./n. ‘id.’; as an ō-stem in Goth. boka f. ‘letter
character’, OFris. b k f./n. ‘book’; as a (j)ōn-stem in OE b ce, b ce f. ‘beech’, OS b ke f.
‘id.’ and OHG buocha f. ‘beech’. To be reconstructed as PIE *bheh2g(-eh2)- or
*bhoh2g(-eh2)- to the root PIE *bheh2g- ‘beech’, cf. Gr. φηγός f. ‘oak’, Gr.(Dor.) φ γός f.
‘id.’ and Lat. f g s f. ‘beech’. Root noun forms are known only in Germanic the
remaining branches displaying something as remarkable as a feminine o-stem. Griepentrog
(1995: 73-74) believes that both forms are archaic and that they have originally both meant
‘beech’ the feminine o-stem originally being an adjective of appurtenance secondarily
altered into a synchronically more transparent feminine ō-stem in Germanic, cf. also
Thöny (2013: 105-106). If Griepentrog is right in his analysis, PG *b k- belongs to layer I.
If, however, as is equally feasible, the Germanic root noun has arisen as a consequence of
the speakers of Proto-Germanic not knowing how to analyse a feminine a-stem and thus
turning it into partly a feminine root noun partly a feminine ō-stem, PG *b k- would
belong to layer IIb. Literature: Bammesberger (1990: 197-198), Bjorvand & Lindeman
(2000: 92-94, 126-127), Boutkan & Siebinga (2005: 55), Casaretto (2004: 43-44),
Griepentrog (1995: 59-77), Holthausen (1974: 28), IEW (2005: 107-108), Kluge/Seebold

11 Hamp (1978: 29-31) thus reconstructs nom.sg. PIE *HaénHat-s (= PIE: *h2énh2t-s) > *ánәt-s > *ánәþ-s > PG *anuþ-
s; nom.pl. PIE *HaénHat-es (= PIE *h2énh2t-es) > *ánәt-es > *ánәþ-es > PG *anþ-iz ~ *and-iz.; and gen.sg. PIE
*HanHat-ós (= PIE *h2nh2t-ós) > *nәt-ós > *̥n̄þ-´/*naþ-´ > *und-/nad-  *an(ә)d- > PG *an(u)d-.

24
(2002: 156-157), Kroonen (2013: 71-72), Lehmann (1986: 77-78), Lloyd et al. (1998: 437-
442, 445-449), Orel (2003: 51-52), Philippa et al. (2003: 289, 339-340), Sehrt (1966: 58),
De Vries (1962: 47-48).
6 PG *b t- f. ‘penalty compensation’. Attested as a root noun only in ON bót f. ‘id.’; as an
ō(n)-stem in the remaining Germanic languages. Belonging to layer III. Literature:
Kroonen (2013: 72), Orel (2003: 52), Wessén (1958: 70).
7 PG *br k- f. ‘trousers breeches’ (mostly pl.). Attested as a root noun in ON br kr f. ‘id.’
(pl.), OE br c f. ‘id.’ (sg.) br c, br c (pl.), OFris. br c f. ‘id.’ (pl.), OS brog(?) f. ‘id.’
(sg./pl.), OHG bruoh, pruoh, pruah f. ‘id.’. According to Griepentrog (1995: 88-90)
borrowed from PCelt. *br k- after the Germanic sound shift but before PIE * > PG * , cf.
Gaul. br c-, br ca, which has probably also been borrowed into the Romance languages,
e.g. Ital. braca f. ‘trousers’ Prov. braya f. ‘id.’ OFrench braie f. ‘id.’ Port. braga f. ‘id.’
etc. Given this rather plausible analysis, PG *br k- belongs to layer IIa.12 Literature:
Bammesberger (1990: 198), Bjorvand & Lindeman (2000: 103-104), Griepentrog (1995:
79-90), Holthausen (1974: 35), IEW (2005: 165), Kluge/Seebold (2002: 153), Kroonen
(2013: 78), Lloyd et al. (1998: 390-394), Orel (2003: 57), Philippa et al. (2003: 384), De
Vries (1962: 58).
8 PG *brust- f. ‘breast chest’. Attested as a root noun or a monosyllabic consonant stem in
Goth. brusts f. ‘id.’ (pl.), OFris. brust, burst f. ‘id.’ and OHG brust, prust f. ‘id.’; a vrddhi
derivative with radical full grade is attested in the a-stem PG *breusta- n. ‘heart courage
breast’ > ON brjóst n. ‘id.’, OE br ost n. ‘id.’ OFris. briast n. ‘id.’, OS briost, breost n.
‘id.’ cf. e.g. Griepentrog (1995: 469-470). To be reconstructed as PIE *bhrus-t- to the root
PIE *bhre s-t- ‘swell’; cf. without the t-enlargement of the root OIr. brú f. ‘abdomen
womb’ bruinne m. ‘breast bosom chest’ etc. Belonging to layer I. Literature:
Bammesberger (1990: 200), Bjorvand & Lindeman (2000: 113-114), Boutkan & Siebinga
(2005: 64-65), Casaretto (2004: 430-431), Griepentrog (1995: 463-471), Holthausen
(1974: 34), IEW (2005: 170-171), Kluge/Seebold (2002: 155), Kroonen (2013: 76, 80),

12 Alternatively PG *br k- is formed with a lengthened o-grade to the PIE root *bhreg- ‘break’ cf. PG *brekana- s.v.
‘break’ and Lat. frang v. ‘break’ s ffr g f. ‘a joint in the hind leg of a quadruped’ (n-stem), cf. the secondary
meaning ‘behind bum’ of PG *br k- which is sparsely attested in West Germanic. In that case early borrowing
could have taken place in the opposite direction, i.e. from Germanic to Celtic.

25
Lehmann (1986: 82), Lloyd et al. (1998: 399-402), Orel (2003: 56, 59), Philippa et al.
(2003: 358), Sehrt (1966: 63), De Vries (1962: 57-58).
9 PG *brū- f. ‘brow’. Attested as a wō-stem in OE brū f. ‘brow’ (only pl.);13 also ON brún f.
‘brow; edge’ (pl. brýnn) which is, in reality, an n-stem with radical as well as suffixal zero
grade. To be reconstructed as PIE *h3bhruH- ‘(eye)brow’, cf. Skt. b rū- f. ‘id.’, Gr. ὀφρύς
f. ‘id.’, Lith. bruvìs f. ‘id.’ OCS brъvь f. ‘id.’ etc. Belonging to layer I. Literature:
Bjorvand & Lindeman (2000: 111-112), Griepentrog (1995: 329-330), Holthausen (1974:
36), IEW (2005: 172-173), Kroonen (2013: 79), NIL (2008: 41-44), Orel (2003: 60), De
Vries (1962: 60).
10 PG *burg- f. ‘city town citadel’. Attested as a root noun in Goth. baurgs f. ‘fortified
place; town’, OE burg, burh f. ‘city’, OFris. burich, burch m. ‘castle; city’, OS burg f. ‘id.’
OHG burg, purk f. ‘id.’; as an ō/i-stem in ON borg f. ‘town; citadel; small hill’. To be
reconstructed as PIE *bhrĝh- to the root PIE *bherĝh- ‘elevate’, cf. also e.g. Av. bәrәz- f.
‘mountain’ adj. ‘high tall’ OIr. brí f. ‘hill elevated spot’ and further Hitt. parku- adj.
‘tall large’ Toch. A pärkär, B pärkäre adj. ‘long’ Skt. b át v. ‘becomes strong’ Arm.
barjr adj. ‘high tall’. Belonging to layer I. Literature: Bammesberger (1990: 63-64),
Bjorvand & Lindeman (2000: 96), Boutkan & Siebinga (2005: 65-66), Casaretto (2004:
39-40), Griepentrog (1995: 91-116), Holthausen (1974: 38), IEW (2005: 140-141),
Kluge/Seebold (2002: 161), Kroonen (2013: 85), Lehmann (1986: 64-65), LIV (2001: 78-
79), Lloyd et al. (1998: 457-460), NIL (2008: 30-34), Orel (2003: 195-196), Sehrt (1966:
66), De Vries (1962: 50).
11 PG *dulþ- f. ‘festival celebration’. Attested as a root noun only in Goth. dulþ f. ’id.’ (only
1x. dat.sg. dulþ); normally as an i-stem in Gothic, cf. e.g. dat.sg. dulþai, and in the
remaining Germanic languages, cf. e.g. OHG tuld f. ‘id.’ (< PG *dulþi-), tult f. ‘id.’ (< PG
*duldi-). According to Griepentrog (1995: 487) maybe misinterpretation of acc.sg. and
thus no basis for assuming root noun inflection at all. Literature: Casaretto (2004: 515-
516), Griepentrog (1995: 487), Kroonen (2013: 108), Lehmann (1986: 97), Orel (2003:
79), Schaffner (2001: 234).

13 Originally maybe OE brūa (nom.pl.) < PG *brū w -, i.e. the dual form of the root noun PG *brū-, cf. e.g.
Griepentrog (1995: 330).

26
12 PG *dur- f. ‘door’ (only pl.). Attested as a root noun in ON dýrr f.pl./n.sg. ‘id.’; as an a-
stem in Goth. daur n. ‘door’ (only sg.), OE dor n. ‘door, gate’ OS dor, dur n. ‘id.’, OHG
tor n. ‘id.’; as an ōn-stem in Goth. daurons f. ‘doors’ (only pl.); as a u-stem transferred
from the root noun inflection in OE duru f. ‘id.’, OFris. dure, dore f. ‘id.’, OS duru f. ‘id.’,
OHG turi f. ‘id.’ (with further developments and transitions). To be compared to the root
noun PIE *dh or- ~ *dhur- ‘(double) door’ (only du.) cf. also e.g. Skt. dv ra (with d
instead of dh from Skt. dvau ‘two’) Arm. dur-kc sb. ‘id.’ (pl.) Gr. θύρα f. ‘id.’, Gr.(Ion)
θύρη f. ‘id.’, Alb. derë f. ‘id.’ Lat. for s f. ‘id.’ (pl.), OIr. dorus n. ‘id.’, Lith. dùrys f. ‘id.’
(pl.). Belonging to layer I. Literature: Bammesberger (1990: 197), Bjorvand & Lindeman
(2000: 171-172), Boutkan & Siebinga (2005: 77), Casaretto (2004: 63), Griepentrog (1995:
117-152), Holthausen (1974: 75, 80), IEW (2005: 278-279), Kluge/Seebold (2002: 935),
Kroonen (2013: 110), Lehmann (1986: 89), NIL (2008: 130-135), Orel (2003: 79-80),
Philippa et al. (2003: 558-559), Sehrt (1966: 88), De Vries (1962: 91).
13 PG *fingr- m. ‘finger’. Attested as a root noun only in ON fingr m. ‘id.’; as an a-stem (PG
*fing-ra-) in the remaining Germanic languages. Belonging to layer III. Literature:
Kroonen (2013: 141), De Vries (1962: 120), Wessén (1958: 69).
14 PG *fl - f. ‘layer stratum’. Attested as a root noun only, if at all, in ON fló, flá f. ‘id.’; as
an ō-stem otherwise. If attested with root noun inflection at all, cf. Griepentrog (1995: 452-
454), this lexeme must belong to layer III. Literature: Griepentrog (1995: 452-454),
Kroonen (2013: 148), Orel (2013: 108), De Vries (1962: 127, 132).
15 PG *f t- m. ‘foot’. Attested as a root noun in ON fótr m. ‘id.’, OE f t m. ‘id.’, OFris. f t m.
‘id.’, OS f t m. ‘id.’, OHG fuoz, fuaz m. ‘id.’ (though normally inflected as an i-stem); as a
u-stem in Goth. fotus m. ‘id.’. To be reconstructed as PIE * d-, i.e. the stem of the
nom.sg. of the root noun PIE * d-s (nom.sg.), * d- (acc.sg.), *péd-s (gen.sg.) ‘foot’,
cf. also e.g. Skt. pad- m. ‘id.’, Av. pad- m. ‘id.’ Gr. πούς (πο -) m. ‘id.’, Arm. otn c. ‘id.’,
Lat. p s (ped-) m. ‘id.’ etc. Belonging to layer I. Literature: Bammesberger (1990: 198),
Bjorvand & Lindeman (2000: 249-251), Boutkan & Siebinga (2005: 136), Casaretto (2004:
42), Griepentrog (1995: 153-183), Holthausen (1974: 113), IEW (2005: 790-792),
Kluge/Seebold (2002: 324), Kroonen (2013: 152), Lehmann (1986: 121), Lloyd & Lühr
(2007: 649-652), NIL (2008: 526-540), Orel (2003: 110), Philippa et al. (2009: 551-552),
Sehrt (1966: 147-148), De Vries (1962: 139).

27
16 PG *fur- adv. ‘for in front of’ etc. with a wide array of descendants in the individual
Germanic languages. To be reconstructed as PIE *pr 2)- ‘across’ etc. Belonging to layer I.
Literature: Griepentrog (1995: 443), IEW (2005: 810-816), Kroonen (2013: 161), Orel
(2003: 119).
17 PG *furh- f. ‘furrow’. Attested as a root noun only in OE furh f. ‘id.’; as an i-stem in OHG
furh f. ‘id.’; as an ō-stem in ON for f. ‘id.’. To be reconstructed as PIE *prok̂- ~ *prk̂-, cf.
the root noun Gr. πρόξ f. ‘roe deer’ πρώξ f. ‘dewdrop’ whose semantic connection to PG
*furh- is extremely weak, though; cf. further Lat. porca f. ‘soil between two furrows’ Celt.
*r c f. ‘furrow’ (e.g. Gallo-Lat. rica) etc. Probably belonging to layer I. Literature:
Bammesberger (1990: 196), Bjorvand & Lindeman (2000: 262-263), Griepentrog (1995:
186-199), Holthausen (1974: 119), IEW (2005: 821), Kluge/Seebold (2002: 322), Kroonen
(2013: 160), Lloyd et al. (1998: 679-681), Orel (2003: 120), Philippa et al. (2009: 566), De
Vries (1962: 137).
18 PG *gait- f. ‘goat’. Attested as a root noun in Goth. gaits f. ‘id.’ (or i-stem), ON geit f.
‘id.’, OE g t f. ‘id.’, OS g t f. ‘id.’; as an i-stem in OHG geiz f. ‘id.’. From PIE *gha d- vel
sim. (original PIE a-vowel), cf. Lat. haedus m. ‘young goat, kid’. Apparent cognates
without an initial consonant and with a different final consonant also exist, cf. e.g. Gr. αἴξ
(αἰγ-) m./f. ‘goat’ and the possible connection to PG *tig n- f. ‘billy-goat’ > OHG ziga f.
‘id.’ and additional roots with similar semantics. Kroonen (2012: 246-247) follows
D’iakonov (1985: 132) and presumes borrowing from a Caucasian language into the
European languages in which case this lexeme would definitely belong to layer IIa;
however, if we assume the possibility of a reconstruction PIE *gha d- with true PIE *a, PG
*gait- could belong to layer I instead. Literature: Bammesberger (1990: 195), Bjorvand &
Lindeman (2000: 291), Casaretto (2004: 40), Griepentrog (1995: 201-209), Holthausen
(1974: 124), IEW (2005: 409-410), Kluge/Seebold (2002: 340), Kroonen (2013: 163-164,
516), Lehmann (1986: 140-141), Orel (2003: 123), Philippa et al. (2005: 207), De Vries
(1962: 162).
19 PG *gans- f. ‘goose’. Attested as a root noun in ON gás f. ‘id.’, OE g s f. ‘id.’; as an i-
stem in OHG gans f. ‘id.’. To be reconstructed as PIE *ĝhans- ‘goose’ (original PIE *a),
cf. also e.g. Skt. a á- m. ‘id.’, Gr.(Att./Ion.) χήν m./f. ‘id.’, Gr.(Dor./Boeot.) χ ν m./f.
‘id.’, Lat. nser m./f. ‘id.’ OIr. géis f. ‘swan’ Lith. žąs s f. ‘goose’ etc. Belonging to layer
I. Literature: Bammesberger (1990: 196), Bjorvand & Lindeman (2000: 333-334),

28
Griepentrog (1995: 211-232), Holthausen (1974: 135), IEW (2005: 412), Kluge/Seebold
(2002: 329), Kroonen (2013: 168), Orel (2003: 126), Philippa et al. (2005: 165-166), De
Vries (1962: 157).
20 PG *gl d- f. ‘red-hot ember’. Attested as a root noun only in ON glóð f. ‘id.’; as an i-stem
in the remaining Germanic languages. Belonging to layer III. Literature: Kroonen (2013:
182), Orel (2003: 137), De Vries (1962: 175), Wessén (1958: 70).
21 PG *hand- f. ‘hand’. Attested as a root noun only in ON ǫnd f. ‘id.’; as a u-stem in the
remaining Germanic languages. Belonging to layer III. Literature: Kroonen (2013: 207-
208), Orel (2003: 159), De Vries (1962: 281), Wessén (1958: 70).
22 PG *hind- f. ‘fallow buck hind’. Attested as a root noun only in ON hind f. ‘id.’; as an
ō(n)-stem in the remaining Germanic languages. Belonging to layer III. Literature:
Kroonen (2013: 226), De Vries (1962: 228), Wessén (1958: 70).
23 PG *hnik- f. ‘sheaf’. Attested as a root noun only in OEN nek f. ‘id.’ (rather uncertain
attestation). According to Nielsen (2000: 299) related to Norw.(dial.) nik n. ‘little cut-off
flakes’. Also the Danish by-form næg from the 16th c. indicates that *(h)nik- be the correct
reconstruction (Lars Brink, p.c.). Brøndum-Nielsen (1935: 163-164) and Griepentrog
(1995: 462), on the other hand, assume PG *naik- on the basis of alleged OEN n k.
Belonging to layer III. Literature: Brøndum-Nielsen (1935: 163-164), Griepentrog (1995:
462), Nielsen (2000: 299).
24 PG *hnit- ~ *gnit- f. ‘nit’. Attested as a root noun in OSw. gnit14 f. ‘id.’ and OE hnitu f.
‘id.’ (acc.pl. hnite); also OHG niz, hniz f. ‘id.’. To be reconstructed as PIE *k̂n d- which
alternates with PIE *k̂on-id- in e.g. Gr. κονίς (κονί -) f. ‘nit’, Alb. thëni f. ‘id.’. Kroonen
(2012: 247) presumes substrate origin of this etymon on the basis of the suffix *-id(h)- ~
*-ind(h)- etc. also found in e.g. PG *arwīt- f. ‘pea’ corresponding to Gr. ἐρέβινθος m.
‘chickpea’. Belonging to layer IIa. Literature: Griepentrog (1995: 474-476), Holthausen
(1974: 166), IEW (2005: 608-609), Kluge/Seebold (2002: 653), Kroonen (2013: 236), Orel
(2003: 180), Philippa et al. (2007: 412).
25 PG *hnut- f. ‘nut’. Attested as a root noun in ON hnot f. ‘id.’, OE hnutu f. ‘id.’ (acc.pl.
hnyte); i-stem in OHG (h)nuz f. ‘id.’. From PIE *knud-; for cognates without the d-
extension cf. e.g. Lat. nux (nuc-) f. ‘nut tree’ OIr. cnú f. ‘nut’. Kroonen (2012: 248)

14 The g in this and related North Germanic forms may originate from PG *ga-hnit- with collectivising prefix PG *ga-.

29
presumes substrate origin on the basis of the widely divergent root extensions: PG *-d- (<
PIE *-t-), Lat. -k- and Celtic vowel length from PIE *-H- (Lat. -k- may result from
laryngeal hardening, though) of which the Germanic one may be described as a substrate
suffix *-u(n)d(h)- (a variant of *-id(h)- ~ *-ind(h)-). Literature: Bjorvand & Lindeman (2000:
676), Griepentrog (1995: 471-473), Holthausen (1974: 167), IEW (2005: 558),
Kluge/Seebold (2002: 658), Kroonen (2013: 237-238), Orel (2003: 181), Philippa et al.
(2007: 433), De Vries (1962: 244-245).
26 PG *idis-/*edis- f. ‘lady’. Attested as a root noun in OE ides f. ‘female lady’ OS idis, ides
f. ‘wife’ OHG itis f. ‘(divine) woman’; as an i-stem in ON dís f. ‘woman girl; fairy,
nymph; goddess’ if we choose to follow Grimm (1844 [1865]: 4-5) and with him Kroonen
(2012: 248-250). Interpreted by Kroonen (2012: 249-250), on the basis of unexpected
vowel alternation between the West and North Germanic examples, as a substrate word
with the substrate prefix *a- (umlauted at an early stage to *e-) in PG *idis-/*edis- f. ‘lady’
which also displays reduction of the radical vowel resulting from the prefigation, the full
vowel being attested in the unprefigated form represented by ON dís. For previous
attempts of establishing an etymology for PG *idis-/*edis- f. ‘lady’ cf. my treatment of this
lexeme in Ph.d. article no. 3. Belonging to layer IIa. Literature: Bjorvand & Lindeman
(2000: 429-430), Holthausen (1974: 2-3, 10), IEW (2005: 11-12, 16-17), Kroonen (2013:
8, 11, 14), Lehmann (1986: 298), Lloyd et al. (1998: 1024-1025), Orel (2003: 6-7, 10),
Sehrt (1966: 94), De Vries (1962: 96, 98-100, 285).
27 PG *kinn- f. ‘cheek’. Attested as a root noun only in ON kinn f. ‘id.’; predominantly as a u-
stem in the remaining Germanic languages. Belonging to layer III. Literature: Kroonen
(2013: 288), Orel (2003: 212), De Vries (1962: 309), Wessén (1958: 70).
28 PG *kl w- f. ‘claw’. Attested as a root noun only in ON kló f. ‘id.’; as an ō-stem (PG
*kl w -) in the remaining Germanic languages. Belonging to layer III. Literature: Orel
(2003: 216), De Vries (1962: 317), Wessén (1958: 70).
29 PG *kwerk- f. ‘throat’. Attested as partly a root noun, partly an i-stem only in ON kverk f.
‘throat’ (pl. ‘neck’); as an ō-stem in the remaining Germanic languages. Belonging to
layer III. Literature: Kroonen (2013: 317), Orel (2003: 228), De Vries (1962: 337), Wessén
(1958: 70).
30 PG *kw - f. ‘cow’. Attested as a root noun in ON kýr f. ‘id.’, OE cū f. ‘id.’, OFris. kū f.
‘id.’, OHG chuo f. ‘id.’ (also as an i-stem); as an i-stem possibly in OS k f. ‘id.’.

30
Hollifield (1979: 54) has explained the Germanic paradigm as *kw - (> *k - ~ *kū-)
abstracted from the acc.sg. PIE *gw of the diphthongal stem PIE *gw - ~ *gwo - vel
sim. ‘cow’, cf. also e.g. Skt. gáv- m./f. ‘id.’, Av. g , Gr. βοῦς m./f. ‘cow bull’, Gr.(Dor.)
βῶς m./f. ‘id.’, Lat. b s (bov-) m./f. ‘cow’, OIr. bó f. ‘id.’ etc. Belonging to layer I.
Literature: Bammesberger (1990: 200), Bjorvand & Lindeman (2000: 482-485),
Griepentrog (1995: 233-256), Holthausen (1974: 62), IEW (2005: 482-483),
Kluge/Seebold (2002: 544), Kroonen (2013: 299), NIL (2008: 189-195), Orel (2003: 219-
220), Philippa et al. (2007: 98), De Vries (1962: 340-341).
31 PG *lūs- f. ‘louse’. Attested as a root noun in ON lús f. ‘id.’, OE lūs f. ‘id.’; i-stem in OS
lūs f. ‘id.’ and OHG lūs f. ‘id.’. To be reconstructed as PIE *luH-s- (with the *s maybe
originating from hypostasis of the nom.sg. ending), cf. also e.g. W llau sb. ‘louse’ (pl.) and
Toch. A lw sb. ‘animal’ B lw sa sb. ‘id.’ (pl.). Belonging to layer I. Literature:
Bammesberger (1990: 195), Bjorvand & Lindeman (2000: 551), Griepentrog (1995: 257-
263), Holthausen (1974: 207), IEW (2005: 692), Kluge/Seebold (2002: 561), Orel (2003:
252), Philippa et al. (2007: 272), De Vries (1962: 369).
32 PG *mann- m. ‘man’. Partial root noun inflection attested in Goth. manna m. ‘id.’ (n-
stem), ON maðr, mannr m. ‘id.’ (a-stem), OE man(n), mon(n) m. ‘id.’ (a-stem), OFris.
mann, monn m. ‘id.’ (a-stem), OS mann m. ‘id.’ (a-stem), OHG mann m. ‘id.’ (a-stem).
Probably to be compared to PIE *mon -o-, i.e. a thematisation of PIE *monu- ‘man’ cf.
e.g. Skt. manu- m. ‘man progenitor’ and the u-stem reflected in WG Mannus. An
alternative etymology suggests that PG *mann- split off from PIE *dhĝhm-on- ‘man’ cf.
e.g. Lat. o m. ‘man’. Whichever be the correct etymology, the transition to root noun
inflection is without doubt a secondary development within Germanic. Belonging to layer
IIb. Literature: Bammesberger (1990: 201), Bjorvand & Lindeman (2000: 584-585),
Boutkan & Siebinga (2005: 266-267), Casaretto (2004: 44-45), Holthausen (1974: 215),
IEW (2005: 700), Kluge/Seebold (2002: 596), Kroonen (2013: 353-354), Lehmann (1986:
244-245), Orel (2003: 260), Philippa et al. (2007: 296-297), Sehrt (1966: 359-362), De
Vries (1962: 374-375).
33 PG *mark- f. ‘border region; mark (unity)’. Attested as a root noun only in ON mǫrk f.
‘mark (unity)’ (partially inflected as an i-stem) and maybe in OFris. merk f. ‘certain
currency’; ō-stem in Goth. marka f. ‘border’, ON mǫrk f. ‘border area forest’ OE mearc f.
‘boundary district’, OS marka f. ‘border region’, OHG marcha f. ‘border region end’.

31
To be reconstructed as PIE *mroĝ- ~ *mrĝ-, cf. the root noun of PCelt. *brog- ~ *brig-
(Gaul. -broges, OIr. brí); secondary i-stem in PCelt. *mrogi- ~ *brogi- (> OIr. mruig,
bruig m. ‘mark landscape’) and o-stem in Av. arәzә m. ‘border mark’ (acc.sg.) etc. In
Germanic terms, probably an original PG a-stem ō-stem or both (PIE *morĝ-o/eh2- > PG
*marka -) with Schwebeablaut that was formed from the zero grade of the root noun, i.e.
PIE *mrĝ- > PG *murk-  PG *mark-a -. Alternatively, PCelt. may have formed *brog-
analogically from the zero grade PCelt. *brig- (< PIE *mrĝ-); consequently, the root noun
would have been PIE *morĝ- ~ *mrĝ-, and the PG form might be a direct continuation of
the PIE root noun. Belonging either to layer I or to layer IIb. Literature: Bjorvand &
Lindeman (2000: 588-589), Boutkan & Siebinga (2005: 258), Casaretto (2004: 45-46),
Griepentrog (1995: 265-285), Holthausen (1974: 217), IEW (2005: 738), Kluge/Seebold
(2002: 599), Kroonen (2013: 355), Lehmann (1986: 246), Orel (2003: 262), Philippa et al.
(2007: 309), Sehrt (1966: 367-368), De Vries (1962: 401).
34 PG *med- adv. ‘with within’ with cognates in all the major individual Germanic
languages. To be reconstructed as PIE *médh-i ‘with within in the middle of’ (loc.sg.).
Belonging to layer I. Literature: Griepentrog (1995: 443) IEW (2005: 702), Kroonen
(2013: 360), Orel (2003: 268).
35 PG *meluk- f. ‘milk’. Attested as a root noun in Goth. miluks f. ‘id.’ (only gen.sg. miluks
1x), ON mjólk, mjolk f. ‘id.’, OE meolc f. ‘id.’, OFris. melok f. ‘id.’, OHG miluch f. ‘id.’;
as an a-stem in OS miluk (only gen.sg. milukas 1x). From PIE *h2 elĝ- (hardly PIE
*h2 elHĝ- > PG *meluk-, cf. Gr. ἀμέλγω f. ‘milk’ without traces of PIE schwa; however
cf. also Lith. lžu (with traces of laryngeal) instead of * el̂žu unless the nature of the
Lith. accent be explained from an original long vowel (Narten verb); for an overview cf.
Hansen (2007: 165)); further cf. Skt. m r v. ‘wipes off’ Av. arәza t , әrәza t v.
‘brushes touches lightly’ Gr. ἀμέλγω v. ‘milk’ Lat. m lge v. ‘id.’ MIr. bligim v. ‘id.’
Lith. lž v. ‘id.’ etc. PG *meluk- may be explained as a contamination of full grade
*melk- and secondary zero grade *mluk- (according to Griepentrog (1995: 300-301 with
further lit.) formed analogically from a secondary full grade PG *mlek-s < *m lk-s in the
nom.sg.). Bammesberger (1990: 197), however, remains sceptical to that analysis and
prefers to explain the *u of PG *meluk- as arisen due to analogical influence from PG
*aluþ- n. ‘ale beer intoxicant’. Belonging to layer I. Literature: Bammesberger (1990:
196-197), Bjorvand & Lindeman (2000: 595-597), Boutkan & Siebinga (2005: 254-255),

32
Casaretto (2004: 40-41), Griepentrog (1995: 287-304), Holthausen (1974: 219), IEW
(2005: 722-723), Kluge/Seebold (2002: 619), Kroonen (2013: 364), Lehmann (1986: 256),
Orel (2003: 267), Philippa et al. (2007: 331-332), De Vries (1962: 389).
36 PG *mūs- ‘mouse; muscle’. Attested as a root noun in ON mús f. ‘mouse biceps’, OE mūs
f. ‘mouse’; as an i-stem in OS mūs f. ‘id.’ OHG mūs f. ‘id.’. To be reconstructed as PIE
*muHs- and compared to e.g. Gr. μῦς m. ‘mouse muscle’, Lat. mūs f. ‘mouse’, OCS ь
f. ‘id.’ etc. Belonging to layer I. Literature: Bammesberger (1990: 195), Bjorvand &
Lindeman (2000: 624-625), Griepentrog (1995: 305-321), Holthausen (1974: 227), IEW
(2005: 752-753), Kluge/Seebold (2002: 607), Kroonen (2013: 379), Orel (2003: 278),
Philippa et al. (2007: 391), De Vries (1962: 396).
37 PG *nagl- m. ‘nail’. Attested as a root noun only in ON nagl m. ‘id.’; as an a-stem (PG
*nag-la-) in the remaining Germanic languages. Belonging to layer III. Literature:
Kroonen (2013: 381), Orel (279), De Vries (1962: 403-404), Wessén (1958: 69).
38 PG *naht- f. ‘night’. Attested as a root noun in Goth. nahts f. ‘id.’, ON nátt, nótt, nǫtt f.
‘id.’, OE neaht, niht f. ‘id.’, OFris. nacht f. ‘id.’ OS naht f. ‘id.’, OHG naht f. ‘id.’. To be
reconstructed as PIE *nokwt- ~ *nekwt- ‘night’, cf. also e.g. Hitt. adv. nekuz ‘(in the)
evening’ (gen.sg.), Gr. νύξ f. ‘night’, Lat. nox f. ‘id.’, Lith. naktìs f. ‘id.’ OCS no tь f. ‘id.’
etc. Belonging to layer I. Literature: Bammesberger (1990: 201-202), Bjorvand &
Lindeman (2000: 643-644), Boutkan & Siebinga (2005: 273-274), Casaretto (2004: 433),
Griepentrog (1995: 476-479), Holthausen (1974: 236), IEW (2005: 762-763),
Kluge/Seebold (2002: 643), Kroonen (2013: 381), Lehmann (1986: 262), Orel (2003: 279-
280), Philippa et al. (2007: 403, Sehrt (1966: 403), De Vries (1962: 405).
39 PG *nas- f. ‘nose’. Root noun inflection not preserved in any Germanic language; attested
as an a-stem in Goth. (wein-)nas adj./m. ‘drunk, drunkard’; as a u-stem in OE nasu, nosu f.
‘nose’ (secondary zero grade formed within PG or later), OFris. nose f. ‘id.’; as an ō-stem
in ON nǫs f. ‘nostrill’, OHG nasa f. ‘nose’ (also ōn-stem is possible for OHG nasa). To be
reconstructed as PIE *n s- ‘nose’ with original and ablauting PIE *a, cf. also Skt. n s f.
‘nose’ (du.), Lat. n r s f. ‘nostrill’ (pl. ‘nose’), Lith. nósis f. ‘nose’ etc. The by-form PG
*nus-, as reflected in OE nosu, may have arisen through contamination of PIE *n s- ‘nose’
and * s ne s- ‘sniff’ cf. Griepentrog (1995: 334-335). Belonging to layer I. Literature:
Bammesberger (1990: 199), Bjorvand & Lindeman (2000: 656-657), Boutkan & Siebinga
(2005: 291-292), Casaretto (2004: 58), Griepentrog (1995: 323-351), Holthausen (1974:

33
232, 238), IEW (2005: 755), Kluge/Seebold (2002: 646), Kroonen (2013: 383), Lehmann
(1986: 400), Orel (2003: 281), Philippa et al. (2007: 419), De Vries (1962: 415).
40 PG *(ga-)naut- m. ‘fellow companion’. Attestation as a root noun possibly (and if so only
partially) in OHG g n z, k n z, kan z m. ‘id.’; otherwise as an a- or an n-stem in OHG, and
as an a-stem in the remaining Germanic languages. The root noun forms are rejected as
scribal errors by Griepentrog (1995: 490-491). Literature: Griepentrog (1995: 490-491).
41 PG *n t- f. ‘large (fishing) net’. Attested as a root noun only in ON nót f. ‘id.’ (also with i-
and ō-stem inflection). Probably a vrddhi derivative PG *n t -  PG *natja- n. ‘(fishing)
net’. Belonging to layer III. Literature: Bjorvand & Lindeman (2000: 669), Orel (2003:
289), De Vries (1962: 412), Wessén (1958: 70).
42 PG *rand- f. ‘border, rim’. Attested as a root noun only in ON rǫnd f. ‘border, rim; shield’;
as an i or an ō-stem in the remaining Germanic languages. Belonging to layer III.
Literature: Orel (2003: 297), De Vries (1962: 458), Wessén (1958: 70).
43 PG *rīk- m. ‘ruler, king’. Attested as a root noun in Goth. reiks m. ‘ruler’. Borrowed into
PG from PCelt. *rīg- (< PIE *h3r ĝ-, cf. also Lat. rex m. ‘king’) as reflected in e.g. OIr. rí
m. ‘id.’ before the Germanic sound shift but after PIE * > PCelt. *ī. Belonging to layer
IIa. Literature: Bammesberger (1990: 199), Bjorvand & Lindeman (2000: 725-726),
Casaretto (2004: 46), Griepentrog (1995: 353-366), IEW (2005: 856), Kroonen (2013:
412-413), Lehmann (1986: 283), Orel (2003: 305).
44 PG *s -ing-(?) f. ‘bed with linen’. Attested as a root noun only in ON sæing, sæng f. ‘id.’,
though cf. also Da.(Jutl.) sæŋ’ә f. ‘bed’ (pl.) with “stød” indicating former
monosyllabicity; as an ō-stem in the remaining EN dialects. Belonging to layer III.
Literature: Brøndum-Nielsen (1935: 165), De Vries (1962: 575).
45 PG *sī-dl- (?) f. ‘herring’. Attested as a root noun only in ON síld f. ‘id.’; as an ō-stem
otherwise. Owing to its rather opaque etymology, PG *sī-dl -(?) may be a lexical
borrowing from some unknown source, i.e. a substrate word. So far, such an assumption
remains speculative seeing that there is nothing in the phonotactics of the form that would
suggest so. Kroonen (2013: 436) tentatively reconstructs an old t-stem PIE nom.sg. *séil-
t, gen.sg. *sil-t-ós, thereby suggesting Du. zeelt c. ‘tench’ to be related to ON síld. Either
belonging to layer I or to layer IIa – or to layer III seeing that root noun inflection is, after
all, found only in Old Norse. Literature: Kroonen (2013: 436), Philippa et al. (2009: 652),
De Vries (1962: 475).

34
46 PG *skrūd- n.? ‘robes’. Attested as a root noun only in OE scrūd n. ‘id.’ and maybe only as
a pseudo root noun 1x or 2x (dat.sg. and acc.pl., respectively) in the Psalter glosses;
normally as an a-stem in OE and in the remaining Germanic languages. According to
Griepentrog (1995: 489-490) no or only scant basis for root noun assumptions. Literature:
Griepentrog (1995: 489-490), Holthausen (1974: 284), Orel (2003: 345).
47 PG *spang- f. ‘spangle’. Attested as a root noun only in ON spǫng f. ‘id.’; as an ō- or an i-
stem in the remaining Germanic languages. Belonging to layer III. Literature: Brøndum-
Nielsen (1935: 166), Orel (2003: 362), De Vries (1962: 540).
48 PG *spurd- m./f. ‘track, course’. Attested as a root noun in Goth. spaurds f. ‘id.’ OE spyrd
m. ‘id.’ OHG spurt m. ‘id.’ (normally inflected as an i-stem). Perfect extra-Germanic
cognates are found in Skt. s rd - f. ‘contest fight’ Av. s әrәd- f. ‘zeal alacrity’; all from
PIE *sprdh- ‘competition’, cf. further the verbal root in e.g. Hitt. art- v. ‘escape, run
away’, Skt. spárdhate v. ‘competes rivals’. Belonging to layer I. Literature:
Bammesberger (1990: 197), Casaretto (2004: 41-42), Griepentrog (1995: 367-379),
Holthausen (1974: 314), IEW (2005: 995-996), Kluge/Seebold (2002: 872), Kroonen
(2013: 470-471), Lehmann (1986: 319), LIV (2001: 580-581), Orel (2003: 367).
49 PG *stad- ~ *stuþ- ~ *stud- f. ‘prop, support, post, pillar’. Consonant stem maybe in OE
studu, stuthu f. ‘id.’ (u-stem otherwise) and ON stoð f. ‘id.’ (pl. steðr, støðr); ablaut
relations uncertain, however. Root noun status rejected by Griepentrog (1995: 490) but
ascertained by Schaffner (2001: 639) who, without mentioning PG *stad-, reconstructs
*stuþ- ~ *stud- on the basis of PIE *sth2u-t-, i.e. the zero grade of PIE *steh2 - ‘stand
firm’ which in turn may be regarded as an extension of PIE *steh2- ‘stand’ cf. also with
laryngeal metathesis, e.g. Skt. st ū - f. ‘pillar’ < PIE *stuh2-. Belonging to layer I.
Literature: Griepentrog (1995: 490), Holthausen (1974: 327), IEW (2005: 1004-1013, esp.
1009), Kroonen (2013: 489), NIL (2008: 637-659), Schaffner (2001: 635-640), De Vries
(1962: 550).
50 PG *stang- f. ‘pole’. Attested as a root noun only in ON stǫng f. ‘id.’; as a (j)ō-stem in the
remaining Germanic languages. Belonging to layer III. Literature: Kroonen (2013: 474),
Orel (2003: 371), De Vries (1962: 559), Wessén (1958: 70).
51 PG *strand- f. ‘border, edge; coast, shore’. Attested as a root noun only in ON strǫnd f.
‘id.’; as an ō-stem in the remaining Germanic languages. Belonging to layer III. Literature:
Kroonen (2013: 482), Orel (2003: 380), De Vries (1962: 555), Wessén (1958: 70).

35
52 PG *sū- f. ‘sow’. Attested as a root noun in ON sýr f. ‘id.’, OHG sū f. ‘id.’. The so-called
”gō-extension” in OE sugu f. ‘id.’ and OHG sugu f. ‘id.’ constitutes, according to Kroonen
(2011: 154-155), in reality the regular development of PG *-uwu- (e.g. PG *-uwum < PIE
*- H- ). From PIE *suH-, cf. also e.g. Gr. ὗς m./f. ‘swine sow boar’, Alb. thi m. ‘pig’,
Lat. sūs m./f. ‘pig sow’ etc. Belonging to layer I. Literature: Bjorvand & Lindeman (2000:
873-874), Griepentrog (1995: 381-392), Holthausen (1974: 329), IEW (2005: 1038),
Kluge/Seebold (2002: 786-787), Kroonen (2013: 490), Orel (2003: 389), Philippa et al.
(2009: 661), De Vries (1962: 574).
53 PG *sulh- f./m. ‘plough’. Attested as a root noun only in OE sulh f./m. ‘id.’; a related a-
stem is found in PG *selha- m. ‘seal i.e. (animal) that drags itself along the ground’ > e.g.
ON selr m. ‘id.’, OE seolh m. ‘id.’ and OHG selah m. ‘id.’. To be reconstructed as PIE
*selk- ~ *slk-, cf. Gr. ἕλκω v. ‘pull’ Gr. ὁλκός m. ‘furrow’ Alb. helq, heq v. ‘pull lead’
Lat. sulcus m. ‘furrow’ etc. Belonging to layer I. Literature: Bammesberger (1990: 197),
Bjorvand & Lindeman (2000: 771), Griepentrog (1995: 393-402), Holthausen (1974: 290,
329), IEW (2005: 901), Kluge/Seebold (2002: 837), Kroonen (2013: 432, 491), De Vries
(1962: 469-470).
54 PG *taihw- f. ‘toe’. Attested as a root noun only in ON tá f. ‘id.’; as an ō-stem (PG
*ta w -) in the remaining Germanic languages. Belonging to layer III. Literature:
Kroonen (2013: 505), Orel (2003: 399), De Vries (1962: 578), Wessén (1958: 70).
55 PG *tang- f. ‘tongs’. Attested as a root noun only in ON tǫng f. ‘id.’; as an ō(n)-stem in the
remaining Germanic languages. Belonging to layer III. Literature: Orel (2003: 401), De
Vries (1962: 604), Wessén (1958: 70).
56 PG *tanþ- ~ *tund- m./f. ‘tooth’. Consonant stem in ON tǫnn f. ‘id.’, OE t þ m. ‘id.’,
OFris. t t m. ‘id.’, OS tand m. ‘id.’ and OHG zan, zand m. ‘id.’; u-stem in Goth. tunþus
m. ‘id.’. From PIE *h1d-ónt-/*h1d-nt- ‘tooth’ (lit. ‘eating’ i.e. present participle of *h1ed-
‘eat’) cf. also Skt. dán (dant-), Gr. ὀ ών (ὀ οντ-), Lat. dens (dent-), Lith. dantìs etc.
Belonging to layer I. Literature: Bammesberger (1990: 202), Bjorvand & Lindeman (2000:
923-926), Boutkan & Siebinga (2005: 412-413), Casaretto (2004: 444), Griepentrog (1995:
479-485), Holthausen (1974: 351), IEW (2005: 289), Kluge/Seebold (2002: 1003),
Kroonen (2013: 509-510), Lehmann (1986: 349), NIL (2008: 208-220, esp. 210, 217), Orel
(2003: 401-402), Philippa et al. (2009: 343), Schaffner (2001: 625-631), Sehrt (1966: 522),
De Vries (1962: 604).

36
57 PG *-t(a)ug- f. ‘ørtug (unity)’. Only attested in East Norse and Gutnish (ODa. ørtugh, ørte,
OSw. ørtugh, ortogh, Gutn. ertaug etc.) with optional root noun inflection in the plural.
Probably belonging to layer III. Literature: Brøndum-Nielsen (1935: 155).
58 PG *tīk- f. ‘bitch’. Attested as a root noun only in ON tík f. ‘id.’; as an ō- or an n-stem in
the remaining Germanic languages. Belonging to layer III. Literature: De Vries (1962:
588), Wessén (1958: 70).
59 PG *traf- f. ‘fringe’. Attested as a root noun only in ON f. trefr ‘fringes’ (pl.) which
coexists with trǫf n. ‘id.’ (pl.) to traf n. ‘scarf’; as an ō-stem in the remaining Germanic
languages. Griepentrog (1995: 462) tentatively suggests that the root noun inflection might
be inherited but dares not conclude anything with certainty due to an expressed lack of
comparative evidence. To be reconstructed as PIE *drop-(eh2-) to the root PIE *drep-
‘pluck cut off’ and compared to Skt. dr í- m. ‘mantle garment’ Gr. ρέπω v. ‘pluck cut
off’. Granted the existence of a root noun PIE *drop- > PG *traf- f. ‘fringe’ we would
need to understand its meaning as ‘what has been cut off’ i.e. an o/e-grade root noun with
resultative semantics. Belonging either to layer I or to layer III. Literature: Griepentrog
(1995: 462), IEW (2005: 211), Kroonen (2013: 520), LIV (2001: 128-129), Orel (2003:
408), De Vries (1962: 596-597).
60 PG *turb- f. ‘turf, peat’. Attested as a root noun in OE turf f. ‘id.’ and maybe OS turf f.?
‘id.’ (if not an a-stem); a-stem in ON torf n. ‘id.’, OFris. turf m. ‘id.’, OS turf m.? ‘id.’,
OHG zurf, zurb m. ‘id.’ (if not a root noun). To be reconstructed as PIE *derbh- ~
*drbh- ‘turn’ secondarily ‘entangle entwine’ vel sim. cf. Skt. darbhá- m. ‘tuft of grass,
bunch of grass’ Toch. A tarp sb. ‘river bank covered by scrub’ etc. Belonging to layer I.
Literature: Bjorvand & Lindeman (2000: 953-954), Boutkan & Siebinga (2005: 415),
Griepentrog (1995: 403-419), IEW (2005: 212), Kluge/Seebold (2002: 921), Kroonen
(2013: 527), Orel (2003: 412-413), Philippa et al. (2009: 436), De Vries (1962: 595).
61 PG *þrū - ‘wooden chock hollow trunk’. Attested as a root noun in ON þró f. ‘trough’,
OE þrū f. ‘trough pipe; chest box’, OS t rū f. ‘fetter’, OHG druoh, drū, thruch f.
‘fetter sling trap’ (primarily inflected as an i-stem). Etymology uncertain: either from PIE
*terH - ~ *truH- ‘wear out tear weaken’ with k-extension or from PIE *trunk-, cf. Lat.
truncus adj. ‘destroyed cut off’. Probably belonging to layer I. Literature: Griepentrog
(1995: 421-429), Holthausen (1974: 370), Orel (2003: 427-428), De Vries (1962: 623).

37
62 PG *(fer-)ud- sb. ‘last year’. Attested as a root noun in ON (í) fjǫrð sb. ‘id.’. From PIE
* et- ~ *ut- ‘year’ cf. the exact parallels of Skt. parút indecl. ‘last year’ Arm. herow adv.
‘id.’, Gr. πέρυσι, Gr.(Dor.) πέρυτι adv. ‘id.’. Attested as a root noun also in Hitt. witt- c.
‘year’; cf. further Gr. ϝέτος, ἔτος n. ‘id.’, Lat. vetus (veter-) adj. ‘old’ etc. Belonging to
layer I. Literature: Bammesberger (1990: 199-200), Bjorvand & Lindeman (2000: 229-
230), Griepentrog (1995: 444-448), IEW (2005: 1175), Kroonen (2013: 137-138), Orel
(2003: 101), De Vries (1962: 124).
63 PG *wand- m. ‘wall’. Attested as a root noun only in ODa. wand m. ‘id.’, cf. also
Da.(Fun.) vær f. ‘id.’ (pl.); as a u-stem in the remaining Germanic languages. Belonging to
layer III. Literature: Brøndum-Nielsen (1935: 153-154). Kluge/Seebold (2002: 971-972).
64 PG *weht- f. ‘thing’. Attested as a root noun only in Goth. waihts f. ‘id.’ – and only 1x.
(acc.pl. waihts); normally inflected as an i-stem in Gothic (gen.sg. waihtais) and in the
remaining Germanic languages. According to Griepentrog (1995: 488-489) not necessarily
an original root noun. He rather regards it either as an original ti-stem with unique and
spontaneous analogy to the root noun inflection (maybe caused by the formal coalescence
of the dat.pl. of root nouns and i-stems) or as an old t-stem for which, however, he can find
no attestations elsewhere. Literature: Bammesberger (1990: 143), Casaretto (2004: 433),
Griepentrog (1995: 488-489), Kroonen (2013: 578, Lehmann (1986: 388-389), Orel (2003:
452-453).
65 PG *wīk- f. ‘creek inlet’. Attested as a root noun only in ON vík f. ‘id.’; as an ō-stem in
the remaining Germanic languages. Belonging to layer III. Literature: Orel (2003: 466), De
Vries (1962: 662), Wessén (1958: 70).
66 PG *wintr- m. ‘winter’. Attested as a root noun only in ON vetr m. ‘id.’; as a u-stem (PG
*wintru-) in the remaining Germanic languages. Belonging to layer III. Literature:
Kroonen (2013: 588), Orel (2003: 455-456), De Vries (1962: 658), Wessén (1958: 69).
67 PG *wl - f. ‘fringe’ < ‘the curly’. Attested as a root noun in OE wl f. ‘id.’, OS wl f.
‘tuft’. Etymology uncertain, but according to Griepentrog (1995: 439) possibly to be
regarded as a nomen rei actae PIE *(H olk̂- ~ *(H lk̂-, cf. Gr. λάχνη f. ‘soft hair’ (< PIE
*(H lk̂-sn-eh2), with metathesis PIE *(H lk̂-s (nom.sg.)  *(H l k̂-s, cf. the partially
parallel situation with PG *meluk- f. ‘milk’ above. Belonging to layer I. Literature:
Griepentrog (1995: 431-441), Holthausen (1974: 403), IEW (2005: 1139-1140).

38
68 PG *wrang- f. ‘frame rib (mar.)’. Attested as a root noun only in ON rǫng f. ‘id.’; as an ō-
stem in the remaining Germanic languages. Belonging to layer III. Literature: Orel (2003:
471), De Vries (1962: 458), Wessén (1958: 70).
69 PG *wr t- f. ‘root’. Attested as a root noun in ON rót f. ‘id.’ which was subsequently
borrowed from Old Norse into late OE r t f. ‘id.’; as an i-stem (PG *wurti- f. ‘medicinal
root herb’) in Goth. waurts f. ‘root’, ON urt f. ‘plant’, OE wyrt f. ‘plant herb’, OS wurt f.
‘id.’ OHG wurz f. ‘root herb’; as an ōn-stem in OHG wurza. To be reconstructed as PIE
* re 2d- ~ * r 2d-, cf. e.g. the ih2-stems Gr.(Myc.) wi-ri-za f. ‘root’, Gr. ῥίζα f. ‘id.’,
Gr.(Aeol.) βῥίζα f. ‘id.’ and Lat. r dīx f. ‘id.’. Vine (1999) has attempted to explain the
aberrant, Greek vocalism by schwa secundum. Probably belonging to layer I. Literature:
Bjorvand & Lindeman (2000: 735-736), Casaretto (2004: 187), Griepentrog (1995: 458-
461), Holthausen (1974: 412), IEW (2005: 1167), Kluge/Seebold (2002: 998-999),
Kroonen (2013: 597), Lehmann (1986: 397), Orel (2003: 473, 476), De Vries (1962: 636).

3. Three layers of Germanic root nouns

3.1. Layer I: Root nouns inherited from Proto-Indo-European

Besides the two main types of root nouns, viz. the one ablauting with ó/é and the one ablauting with
é/Ø, we find a third type which actually constitutes a subtype of the ó/é-ablauting root nouns. If, as
Schindler actually pointed out himself (1972: 34-36), an originally ó/é-ablauting root noun has the
root structure -ERT, the weak form would appear with radical zero grade. In other words, we would
obtain a new type with ó/Ø ablaut.
Building on this claim by Schindler, Nielsen Whitehead (2010; 2013) has recently contributed
with a considerable amount of data to the analysis of the ablaut conditionings in the root noun
category. Consequently, she has discovered that, out of 32 examined roots of the structures -RC and
-HC, the vast majority displays general zero grade in both the strong and the weak cases. Further,
she has suggested a rationale for this phenomenon, viz. that root nouns in Proto-Indo-European
eventually come to appear in a form where three criteria are fulfilled, viz. (1) that the root has to
contain at least one consonant in the syllable onset, (2) that a vocalic element must be displayed in
the root, and (3) that no more than one consonant is allowed in the radical syllable coda. The third
constraint, however, does not apply to roots that have two non-vocalisable consonants, i.e. two
obstruents, in their syllable coda.

39
When scrutinising and focusing on the Germanic root nouns, I cannot help reaching the
conclusion that Nielsen Whitehead’s claim is at least partially valid. The root structure indeed
seems to form the decisive factor regarding the expected ablaut grade of a Proto-Germanic root
noun, cf. also Griepentrog (1995: 419) for a similar statement. In particular, I believe inherited
Germanic root nouns to display:

1 Radical full-, o-, or lengthened grade with the root structure (C)CVT(C), (C)CVH(C), CVC:
(5) PG *b k- f. ‘beech; book’ (layer IIb also possible), (15) PG *f t- m. ‘foot’, (30) PG
*kw - f. ‘cow’, (38) PG *naht- f. ‘night’, (41) PG *n t- f. ‘large (fishing) net’ (layer III
also possible), (59) PG *traf- ‘fringe’ (layer III also possible) (67) PG *wl - f. ‘fringe’ <
‘the curly’ and (69) PG *wr t- f. ‘root’ (layer III also possible).
2 Radical zero grade with the root structure (C)CVRC, cf. e.g. (8) PG *brust- f. ‘breast
chest’, (9) PG *brū- f. ‘brow’, (10) PG *burg- f. ‘city town citadel’, (12) PG *dur- f.
‘door’, (17) PG *furh- f. ‘furrow’, (31) PG *lūs- f. ‘louse’, (36) PG *mūs- ‘mouse;
muscle’, (48) PG *spurd- m./f. ‘track course’, (52) PG *sū- f. ‘sow’, (53) PG *sulh- f./m.
‘plough’, (60) PG *turb- f. ‘turf peat’ and (61) PG *þrū - ‘wooden chock hollow trunk’.
3 Preservation of original a-vowel regardless of the root structure: (1) PG *aik- f. ‘oak’, (2)
PG *alh- m. ‘temple sanctuary’ (layer IIa or IIb also possible), (18) PG *gait- f. ‘goat’
(layer IIa also possible), (19) PG *gans- f. ‘goose’ and (39) PG *nas- f. ‘nose’. Without
doubt, many of these may belong to layer IIa rather than to layer I.15

3.1.1. Apparent counterexamples

As neat as this distribution may seem, we cannot avoid considering a range of apparent
counterarguments or exceptions, the first of which is (1) PG *aik- f. ‘oak’ where we would a priori
expect PIE †h2 ĝ- > PG †ik. However, this lexeme only constitutes an exception if we choose to
reconstruct PIE *h2e ĝ- or *h2o ĝ- rather than *a ĝ- with an original (post-)PIE *a in which case PG
*aik- could be analysed as a lexical borrowing. We might at least consider one more alternative,
though, viz. that Germanic could have followed Greek in displaying vocalised reflexes of an initial
laryngeal (at least of PIE *h2) followed by one of the approximants *i or *u, cf. Ph.D. article no. 3,
in which case it would seem possible that PIE *h2 ĝ- > PG *aik- by regular soundlaw.

15 Needless to say, the argument of attributing (some of) these root nouns to layer II rather than layer I gains further
strength if one believes in the theory that Proto-Indo-European had no true *a.

40
(2) PG *alh- m. ‘temple sanctuary’ seems to constitute another exception. It is important to note
however, that we only find root noun inflection of this lexeme in Gothic; in the remaining Germanic
languages, we find an a-stem. Outside Germanic, we find root noun inflection in Greek (only
attested in dat.sg.), too; however cf. the somewhat aberrant semantics. If we are, indeed, dealing
with an inherited root noun rather than e.g. a PG a-stem, in which case PG *alh- would belong to
layer IIb instead, we may be facing an original a-vowel, a possibility that cannot be contradicted by
any other Indo-European language unless relation to the root PIE *h2lek̂s- ‘ward off guard protect’
is invoked, cf. Skt. rák at , Arm. aracel, Gr. ἀλέξω etc. (IEW 2005: 32). The possibility that PG
*alh- is a lexical borrowing and thus is to be affiliated with layer IIa may also be considered, cf. in
this regard Kroonen (2013: 22).
Adverbial or prepositional material such as (3) PG *and-, *anþ-, *und-, *unþ-, *umbi prep.
‘about’, (16) PG *fur- adv. ‘for in front of’ and (34) PG *med- adv. ‘with within’ cannot be said to
be real counterexamples since they all appear only as archaisms in fossilised case forms and
functioning as adverbs or prefixes even in the earliest attestations of Germanic, i.e. prior to the
introduction of the distributional rules governing what ablaut grade to expect in the root of a given
root noun.16 Consequently, these forms may be straightforwardly disregarded.
At a first glance, the most serious counterexample is provided by (33) PG *mark- f. ‘border
region; mark (unity)’ where we would definitely expect †murk-. However, it is far from unlikely
that we are faced with an original ō-stem PG * ark - (or a-stem *marka-) whose secondary full
grade has been formed from the zero grade of the original root noun PIE * roĝ- * rĝ- as attested
in Celtic. In Germanic terms, such a Schwebe-ablaut analysis seems quite plausible seeing that the
zero grade PIE * rĝ- would be vocalised as PG **murk- with the epenthetic vowel inserted before
the resonant. Given that only the zero grade of the original root noun would have survived in
Germanic, the plausibility of a new full grade along the lines of PG *mark- would be greater than
that of PG **mrak-. However, this analysis, too, is far from being unproblematic. First, the
secondary zero grade may have been formed already at the PIE stage, cf. Av. arәzә m. ‘border,
mark’ < PIE * orĝ-o-. Second, PCelt. *brog- may just as easily be viewed as secondarily formed in
analogy with the zero grade PCelt *brig- < PIE * rĝ-, cf. also Schindler (1972: 34-35), in which

16 Neither PG *fur and *med nor PG *unþ-, und- (incl. *umbi), i.e. the by-forms of *anþ-/and-, may reasonably be
regarded as counterexamples at all in that they display the outcome expected from the distributional rules regarding
the root nouns of layer I. Consequently, they are mentioned together with the true counterexample of PG *anþ-,
*and- only for methodological reasons.

41
case the root noun would have been PIE * orĝ- ~ * rĝ- and PG *mark- would reflect the PIE root
noun directly. When accepting that interpretation, we would have to accept PG *mark- as a true and
inexplicable exception to the distributional rules outlined in this article.
If Griepentrog (1995: 300-301 with literature) is right in his analysis of (35) PG *meluk- f. ‘milk’
as a contamination of full grade *melk- and secondary zero grade *mluk- formed from a secondary
full grade PG *mlek-s < * lk-s in the nom.sg., cf. also the discussion in section 2, and if this
contamination took place prior to the generalisation of the radical ablaut grade in root nouns
according to their phonotactics, the possibility arises that the language users have not been able to
analyse and interpret PG *meluk- as a real root noun at the time when the generalisation rules were
applied. Granted this analysis and provided that (4) PG *anad-, *anid-, *anud- f. ‘duck’ actually is
an archaic root noun, the same may account for its failure to display radical zero grade. Its second
vowel, whether arisen from PIE *H or not, may simply have caused it to be analysed as a lexeme
consisting of a root and a suffix rather than a root noun.
An original t-stem and thus not a root noun may be what underlies (49) PG *stad- ~ *stuþ- ~
*stud- f. ‘prop support post pillar’, i.e. PIE *stә2-t- > PG *stad- that was reinterpreted as an o-
grade form to which a new secondary zero grade *stud- was formed in analogy with the ablaut
pattern known from e.g. (56) PG *tanþ- ~ *tund- m./f. ‘tooth’. An even less clear case of a
counterexample emerges if we choose to follow Schaffner’s (2001: 639) analysis of PG *stuþ-/stud-
as a true, and in Germanic terms expected, zero grade formation PIE *sth2u-t-. Consequently,
regardless of the chosen analysis, PG *stud- should not be counted among the valid
counterexamples to the distributional rules outlined in this article. A somewhat similar explanation,
viz. an ablauting t-stem that was still analysable to the language users, might be provided for (45)
PG *sī-dl- (?) f. ‘herring’ if Kroonen (2013: 436) is right in his hypothesis and we are not dealing
with a substrate term.
(56) PG *tanþ- m./f. ‘tooth’ (next to the expected form PG *tund-) might actually count as a
counterexample per se. Being in fact a substantivised present participle (< PIE *h1d-ónt- ~ *h1d-nt-
’eating’) rather than a root noun PG *tanþ- cannot be expected to follow the distributional rules
pertaining to root nouns, at least not if it was still interpreted as a participle by the speakers of (Pre-)
Proto-Germanic, cf. also the retention of ablaut.
The last item on our list of apparent counterexamples is (62) PG *(fer-)ud- sb. ‘last year’ which
should ideally have had full, lengthened or o-grade according to our distributional rules that predict
PG †(fer)wed- vel sim. to be the expected outcome in that all root nouns should have at least one

42
consonant in the syllable onset. The deviation can be explained quite straightforwardly, though,
seeing that PG *ud- never occurs as a simplex but only as a second member of a compound where
radical zero grade of a root noun is expected whenever phonetically possible, cf. e.g. Skt. ratīca
‘face’ (gen.sg.) < PIE *proti-h3kw-os.
A considerable number of the lexemes analysed in section 2 cannot reasonably be treated as
counterexamples even if their radical ablaut grade deviates from what is expected according to our
distributional rules, e.g. (55) “PG” *tang- f. ‘tongs’. The reason for that is that the lexemes in
question had not yet entered the root noun declension at the stage when our distributional rules were
still active. Rather, they were adopted into the root declension at one of the later stages represented
by my suggested layers IIa, IIb or III.

3.2. Layer IIa: Unsegmentable substrate or loan words reanalysed as root nouns in Proto-Germanic

In a recent article, Kroonen (2012: 242-255) gives a convincing account of some problematic
lexemes by listing a number of already known substrate markers, including the substrate suffix PG
*-īt- alternating with *-ūt- whose origin seems to be identical to that of Gr. -ιν - ~ -ινθ- ~ - - ~ - θ-,
the substrate prefix *a-/*e- alternating with *Ø-, and consonant clusters that violate the general
Proto-Indo-European and Proto-Germanic phonotactic constraints. These and further criteria
Kroonen believes to see present in the four root nouns (18) PG *gait- f. ‘goat’ (24) PG *hnit- ~
*gnit- f. ‘nit’ (25) PG *hnut- f. ‘nut’ and (26) PG *idis-/*edis- f. ‘lady’. A common denominator of
these items is, according to Kroonen, that they appeared unsegmentable to the speakers of (Pre-)
Proto-Germanic at the time of the borrowing and were consequently assigned to the root noun
declension. Kroonen also includes the two cases of PG *arwīt- f. ‘pea’ and *wīs nd- ~ *wīz nd- m.
‘European bison Bison bonasus’ to his article but seeing that these are not monosyllabic root
nouns or consonant stems, I have left them out of consideration for the present purpose.
It lies far outside the scope of this article to confirm or reject the details of Kroonen’s
etymologies and general idea.17 However, it can hardly be questioned that a model such as that of
Kroonen, which identifies borrowed root nouns, is of great utility to my claim regarding the
predictability of the ablaut grade of the inherited root nouns (layer I). In other words, if we may
assume that the items on Kroonen’s list were borrowed into (Pre-)Proto-Germanic at a time

17 It should be noted though that Kroonen’s idea has the great general advantage to a range of other substrate theories
that etymologies are proposed on the basis of structural considerations and, in many cases, known possible substrate
sources (rather than with virtually all-embracing “Schallwurzeln” and/or in want of a good intra-lingual etymology).

43
subsequent to the active application of the distributional rules outlined in section 3.1, a root noun
such a PG *gait- and unsegmentable consonant stems such as PG *arwīt-, *edis- and *wīsund-
/wīzund-, whose ablaut grade and, in the latter cases, general root structure would violate these
rules, may easily be disregarded.18 They may be described as later borrowings that took place after
the rules had ceased to be effective.
It should be noted, though, that besides the items listed by Kroonen (2012), yet another handful
of obvious candidates for membership in layer IIa must be considered, viz. (1) PG *aik- f. ‘oak’
(layer I also possible), (2) PG *alh- m. ‘temple sanctuary’ (layer I or IIb also possible), (7) PG
*br k- f. ‘trousers breeches’ (43) PG *rīk- m. ‘ruler king’ and (45) PG *sī-dl- (?) f. ‘herring’
(layer I and III also possible) which are all, with divergent degrees of certainty, possible lexical
borrowings, cf. the discussion in section 2.

3.3. Layer IIb: Nouns from other declensions reanalysed as root nouns in Proto-Germanic or in parts
of the Germanic dialect continuum

Of the three layers outlined in this article, layer IIb is, without doubt, the one most difficult to
define, but we are aided by the descriptions of transitional tendencies and inflectional class profiles
provided by Thöny (2013: 79-82, 314-325). Principally, we may here ascribe nouns that have been
transferred from other declensions (a- and ō-stems) into the root noun declension in the North-West
Germanic language(s) after the split-up of Proto-Germanic, since they are attested with root noun
inflection only (in parts of) Germanic and nowhere else. Furthermore, it is highly likely that the
transitions took place already in (Pre-)Proto-Germanic times with the result that layer IIa and IIb
are, in principle, identical, the only difference being that they are fed with material from two
different sources. At least, PG *mann- (see below) would suggest this latter scenario since it
displays partial root noun inflection not only in North-West Germanic but also in Gothic.
Judging from my etymological considerations in section 2, the following root nouns may fall
under this category: (2) PG *alh- m. ‘temple sanctuary’ (layer I or IIa also possible), (5) PG *b k-
f. ‘beech; book’ (layer I also possible), (32) PG *mann- m. ‘man’ and (33) PG *mark- f. ‘border
region; mark (unity)’ (layer I also possible, cf. the lengthy discussion in section 3.1.1). To these
may be added that, in reality, we cannot know if at least some of the root nouns that do fit the

18 That zero grade is what I would structurally expect in PG *hnit- and *hnut-, even had they been inherited root nouns
(layer I), obviously does not impede my analysis of them as rather being loan words that entered the language at a
later stage.

44
distributional rules outlined in section 3.1 could not just as well have entered the Proto-Germanic
language at this later stage. Evidently it is often difficult to decide with certainty if an item belongs
to layer I or to layer IIb. We can only substantiate our decision by analysing the prevalence of a
given root noun candidate in not only the Germanic but in all Indo-European languages. If a noun is
found with (partial) root noun inflection only within (parts of) Germanic, the likelihood of its proper
attribution to layer IIb rather than to layer I increases considerably.
A further complicating factor regarding this layer is the seeming lack of transparency as to what
phonological, morphological, semantic and other factors catalysed the transition of a noun from one
of the vocalic declensions to the root noun declension. Partial case syncretism in the (North-West)
Germanic languages between the providing a- and ō-stem declensions and the receiving root noun
declension constitutes the, in my view, intuitively most appealing explanation for the declensional
transitions of this layer. The question still remains, however, why so few such nouns have (or could
have) changed their declensional affiliation with the result that they partially or entirely became root
nouns and, consequently, why exactly these approximately four nouns and not, say, PG *geb - f.
‘gift’ ( †geb-) or *fugla- m. ‘bird’ ( †fugl-) were (or could have been) affected.

3.4. Layer III: Nouns from other declensions reanalysed as root nouns in North Germanic

As already mentioned, it is a long-known fact that in North Germanic, the root noun declension
was, at some point, revitalised with the result of it accepting new members from other declensions,
cf. e.g. Brøndum-Nielsen (1935: 146, 154-155). Hence follows that, contrarily to the situation in
layer IIb, North Germanic has had a particularly extensive influx of nouns to the root noun
declension from other declensions. When also taking into account that the catalysts behind this
purely North Germanic development are far more transparent than those behind the similar
developments of layer IIb, we can easily defend the assumption of a fourth layer distinct from and
considerably younger than the former one.19
Two possible catalysts for the transition of masculine nouns are (1) interactions between root
nouns and u-stems owing to partial case syncretism, cf. the opposite analogy in Goth. fotus ‘foot’
(original root noun) and (2) “body part analogy” with ON fótr m. ‘foot’ (< PG *f t-) as the model

19 It can hardly be excluded, though, that this third layer is a mere continuation of the second layer, i.e. that the
developments and mechanisms initiated in layer IIb soon ceased to be productive in West Germanic but continued to
be so in North Germanic where they were even intensified, thus creating what I have here labelled layer III.

45
example. The PG transponates (63) *wand- m. ‘wall’ and (66) *wintr- m. ‘winter’ belong to the
former group; (13) *fingr- m. ‘finger’ and (37) *nagl- m. ‘nail’ (both original a-stems) to the latter.
For the feminines, Brøndum-Nielsen (1935: 138-140, 154-155) lists not only two but three
possible catalysts, the first of which being that u-mutation in the acc.sg. (PIE *- > PG *-un > ON -
Øu) and secondarily in the nom.sg. creates a parallel to the feminine ō-stems (PG *- # > NWG -u#).
Not surprisingly therefore ō-stem nouns with visible u-mutation, i.e. nouns with ON a or á/ó in the
root, are particularly prone to making the declensional transition.20 As such, (4) ON ǫnd, ǫnð f.
‘duck’ (6) bót f. ‘penalty compensation’ (20) glóð f. ‘red-hot ember’ (41) nót f. ‘large (fishing)
net’ (layer I also possible) (42) rǫnd f. ‘edge’ (68) rǫng f. ‘frame rib (mar.)’ (< PG transponate
*wrang-), (69) rót f. ‘root’ (layer I also possible) (47) s ǫng f. ‘spangle’ (50) stǫng f. ‘pole’ (51)
strǫnd f. ‘beach’ (55) tǫng f. ‘tongs’ and (59) trǫf f. ‘fringe’ (layer I also possible) have
transcended to partial root noun inflection. Analogy between the i-mutation in root nouns (gen.sg.
and nom./acc.pl.) and the R-mutation in nouns ending synchronically in a vowel constitutes yet
another catalyst; thus e.g. (14) ON fló, flá f. ‘layer stratum’ (28) kló f. ‘claw’ and (54) tá f. ‘toe’. A
factor that might have encouraged this development, too, is the circumstance that a couple of
inherited root nouns, viz. (30) PG *kw - ‘cow’ (ON kýr) and (52) *sū- ‘sow’ (ON sýr), actually
already had a vocalic auslaut in Proto-Germanic. The third and last easily observable catalyst is the
apparent need for total elimination of the feminine u-stem declension originally represented by (21)
ON ǫnd21 ‘hand’ and (27) kinn ‘cheek’ cf. also the remark above regarding partial case syncretism
between root nouns and u-stems.
A few productively created root nouns of layer III are not that easily explained, however. As
such, it remains enigmatic what catalysts are responsible for the transition of (22) ON hind f.
‘fallow buck hind’ (29) kverk f. ‘throat’ (pl. ‘neck’) (45) síld f. ‘herring’ (44) sæ(i)ng f. ‘bed’

20 Most Old Norse vowels only display a separate u-mutated variant with the w-umlaut; not with the u-umlaut proper.
Only with a, whose u-mutated variant is ǫ (cf. e.g. NWG *hab(a)nu > ON höfn ‘harbour’) and with á, whose u-
mutated variant was (> á, cf. e.g. NWG *sk l > Early ON *skol > ON skál ‘bowl’) when not nasalised and (>
ó/á, cf. e.g. NWG *ansuR > Early ON * sR > ON óss, áss ‘a Norse god’) when nasalised was a difference between
the mutated and non-mutated variants made explicit, cf. e.g. Andersen (1962: 5, 17). Not all instances of ON ó have
developed from u-mutated, nasalised á. In the case of bót, for example, one would have to assume that the transition
from the ō-stem declension to the root noun declension happened relatively late, i.e. at a time when the language
users could no longer distinguish between ON ó < PG * and ON ó < PG * (by u-umlaut and nasalisation).
21 Dat.sg. hendi < PG *handiu < PIE *- , cf. Ph.d. article no. 3.

46
(58) tík f. ‘bitch’ (65) vík f. ‘creek inlet’ (57) OEN/Gutn. *-t(a)ug f. ‘ørtug (unity)’ and possibly
(23) OEN nek f. ‘sheaf’ to the root noun declension.

4. Conclusion
In this article, I have argued that root nouns in Proto-Germanic and the individual Germanic
languages may be attributed to three chronologically defined layers:

1 Layer I: Root nouns inherited from Proto-Indo-European.

2 Layer IIa: Unsegmentable substrate or loan words reanalysed as root nouns in Proto-
Germanic.
3 Layer IIb: Nouns from other declensions reanalysed as root nouns in Proto-Germanic or in
parts of the Germanic dialect continuum.
4 Layer III Nouns from other declensions reanalysed as root nouns in North Germanic.

I have further argued that the ablaut grade of the inherited root nouns (layer I) is predictable from
the phonotactics of the root in partial accordance with the rules suggested by Nielsen Whitehead
(2010; 2013):

1 Radical full-, o- or lengthened grade with the root structure (C)CVT(C), (C)CVH(C), CVC.
2 Radical zero grade with the root structure (C)CVRC.
3 Preservation of original a-vowel regardless of the root structure.

However, these rules should not be applied on the Germanic material until the subsequently created
root nouns of layers IIa, IIb and III have been identified and filtered out. For the sake of clarity, the
considerations made throughout the discussion of the material as to what layer(s) a given root noun
could or should be attributed to, will be summarised in table 1.

Table 1: Attribution of the investigated root nouns to the three layers

Layer I Layer IIa Layer IIb Layer III

(1) PG *aik- f. ‘oak’ X X

(2) PG *alh- m. ‘temple sanctuary’ X X X

47
(3) PG *and-, *anþ-, *und-, *unþ-, *umbi (X)
prep. ‘about’

(4) PG *anad-, *anid-, *anud- f. ‘duck’ X X

(5) PG *b k- f. ‘beech; book’ X X

(6) PG *b t- f. ‘penalty compensation’ X

(7) PG *br k- f. ‘trousers breeches’ X

(8) PG *brust- f. ‘breast chest’ X

(9) PG *brū- f. ‘brow’ X

(10) PG *burg- f. ‘city town citadel’ X

(11) PG *dulþ- f. ‘festival celebration’

(12) PG *dur- f. ‘door’ X

(13) PG *fingr- m. ‘finger’ Х

(14) PG *fl - f. ‘layer stratum’ Х

(15) PG *f t- m. ‘foot’ Х

(16) PG *fur- adv. ‘for in front of’ (Х)

(17) PG *furh- f. ‘furrow’ X

(18) PG *gait- f. ‘goat’ X X

(19) PG *gans- f. ‘goose’ X

(20) PG *gl d- f. ‘red-hot ember’ X

(21) PG *hand- f. ‘hand’ X

(22) PG *hind- f. ‘fallow buck hind’ X

(23) PG *hnik- f. ‘sheaf’ X

(24) PG *hnit- ~ *gnit- f. ‘nit’ X

(25) PG *hnut- f. ‘nut’ X

(26) PG *idis-/*edis- f. ‘lady’ X

(27) PG *kinn- f. ‘cheek’ X

(28) PG *kl w- f. ‘claw’ X

(29) PG *kwerk- f. ‘throat’ X

(30) PG *kw - f. ‘cow’ X

48
(31) PG *lūs- f. ‘louse’ X

(32) PG *mann- m. ‘man’ X

(33) PG *mark- f. ‘border, region; mark X X

(unity)’

(34) PG *med- adv. ‘with within’ (X)

(35) PG *meluk- f. ‘milk’ X

(36) PG *mūs- ‘mouse; muscle’ X

(37) PG *nagl- m. ‘nail’ X

(38) PG *naht- f. ‘night’ X

(39) PG *nas- f. ‘nose’ X

(40) PG *(ga-)naut- m. ‘fellow

companion’

(41) PG *n t- f. ‘large (fishing) net’ X X

(42) PG *rand- f. ‘border rim’ X

(43) PG *rīk- m. ‘ruler king’ X

(44) PG *s -ing-(?) f. ‘bed with linen’ X

(45) PG *sī-dl- (?) f. ‘herring’ X X X

(46) PG *skrūd- n.? ‘robes’

(47) PG *spang- f. ‘spangle’ X

(48) PG *spurd- m./f. ‘track course’ X

(49) PG *stad- ~ *stuþ- ~ *stud- f. ‘prop X

support post pillar’

(50) PG *stang- f. ‘pole’ X

(51) PG *strand- f. ‘border edge; coast X

shore’

(52) PG *sū- f. ‘sow’ X

(53) PG *sulh- f./m. ‘plough’ X

(54) PG *taihw- f. ‘toe’ X

(55) PG *tang- f. ‘tongs’ X

(56) PG *tanþ- ~ *tund- m./f. ‘tooth’ (X)

49
(57) PG *-t(a)ug- ‘ørtug (unity)’ X

(58) PG *tīk- f. ‘bitch’ X

(59) PG *traf- ‘fringe’ X X

(60) PG *turb- f. ‘turf peat’ X

(61) PG *þrū - ‘wooden chock hollow X

trunk’

(62) PG *(fer-)ud- sb. ‘last year’ X

(63) PG *wand- m. ‘wall’ X

(64) PG *weht- f. ‘thing’

(65) PG *wīk- f. ‘creek inlet’ X

(66) PG *wintr- m. ‘winter’ X

(67) PG *wl - f. ‘fringe’ < ‘the curly’ X

(68) PG *wrang- f. ‘frame rib (mar.)’ X

(69) PG *wr t- f. ‘root’ X X

Not all challenges regarding root nouns in Germanic have been solved. At least three problems
remain, viz. (1) that a distinction between an original, Proto-Indo-European a-vowel (layer I) and an
a-vowel that has entered the language through borrowing (layer IIa) is often virtually impossible to
make, (2) that no obvious candidate for a catalyst regarding the transitions of layer IIb can be
defined, and (3) that even in layer III, there is a residual quantity of root nouns where no candidate
for a catalyst is obvious. Future studies will hopefully bring elucidation to some of these matters.

50
Article no. 2: The structure, form and function of the
Germanic primary i-stems

In order to separate archaisms from innovations, we must first define in every

concrete case what characterises an archaic form. In this article, I define the
structure, form and function of archaic, primary i-stems in Germanic. It turns out
that, as has already been established by previous scholarship, primary i-stem verbal
abstracts are synchronically and productively formed with the same ablaut grade as
is found in the stem of the preterite participle of the corresponding strong verb, this
system (together with, i.a., the reanalysis of old s-stems as i-stems) giving rise to a
wide array of new possible Germanic ablaut grades in the root syllable. The only
archaic i-stems are those displaying radical o- and zero grade. Radical o- and zero
grade of i-stems are also found in other Indo-European branches, as are the two main
functional categories of i-stems, viz. adjectival agent nouns on the one hand and
action nouns or verbal abstracts on the other. I further propose that the two formal
types, i.e. the zero grade type and the o-grade type, were originally identical, the
radical phonotactics constituting the determining factor for when to expect zero
grade and when to expect o-grade. Such a complementary distribution of o- and
zero-grade is reminiscent of what Rasmussen (1989: 158-175) suggests for the
Proto-Indo-European eh2-stems of the type Lat. toga/fuga of which these i-stems
should probably be seen as a variant, i.e. adjectival agent nouns of the structure PIE
*CóC-i- or *C´C-i- and action nouns or verbal abstracts of the structure PIE *CoC-í-
or *CC-í-. When transferred to Germanic, the original distribution of o- and zero
grade was abandoned in favour of a new ablaut distribution dependent on the ablaut
system of the strong verbs, and the old system is only recognisable through an, albeit
large, handful of i-stems displaying aberrant ablaut grades in comparison with the
strong verbs corresponding to them.

1. General characteristics of i-stems and their inflection

A considerable number of nouns in Proto-Germanic are inflected as i-stems. In other words, these
nouns are characterised by the addition of a nominal suffix PG *-i- to the lexical root. At least as an
overall type, the i-stems are of Proto-Indo-European pedigree, for which reason a wide array of
inflectional features have also been thus inherited. For inflectional details on the i-stems as well as
on the, in most regards, completely parallel u-stems cf. the detailed treatment in Ph.D. article no. 4.
Suffice it therefore here to list the i-stem case endings of the individual Germanic languages

51
supplemented by their assumed Proto-Germanic and Proto-Indo-European precursors as
reconstructed by, e.g., Krahe (1967: 25-31), Bammesberger (1990: 124-127), Boutkan (1995: 236-
250) and most recently Ph.D. article no. 4.

Table 1: Inflectional endings of the masculine and feminine i-stems

PIE PG Goth. RN ON OE OS OHG

nom.sg. *-i-s *-iz -s -iR -r -e / -Ø -i / -Ø -i / -Ø

acc.sg. *-i-m *-in -Ø -i -Ø -Ø -Ø -Ø

gen.sg. *- i-s *-aiz -ais - -ar -æ/-e or  dat.sg.  dat.sg.

 dat.sg.

dat.sg. *- i-ei *-ai -ai - -ai -i -æ/-e - -

loc.sg. *- i *-ei > *-ī - - - -i D -i D -i D

*- i *-ai -ai D22 - -ai D -i D -æ/-e D - -

instr.sg. *-i-h1 *-ī - - - -i/-e -i -

voc.sg. *- i *-ai = nom.sg. = nom.sg. = nom.sg. = nom.sg. = nom.sg.

*-i ? *-i ? -Ø

nom.pl. *- i-es *-īz -eis - -ir -e -i -i

acc.pl. *-i-ns *-inz -ins - -i -e -i -i

gen.pl. *-i-ōm  *-jaōn ? -e - - - -io -io

instr.pl. *-i-bhis *-imiz -im - - - -in -im

A considerable number of forms are lacking in table 1. Except for Runic Norse which is so poorly
attested that examples of several case forms are simply not known at all, the lacunae are mainly
caused by a given case form being replaced by another case form either by means of case
syncretism internally within the i-stem paradigm, e.g. between dat.sg. and loc.sg., or by means of
transition of a desinence from another inflectional class into the i-stems, e.g. the i-stem instr.pl. ON
-um, OE -um originating in the a-stem inflection. Such transitions happened not only when a case
form was completely replaced; in several cases the inherited i-stem desinences coexist with new
ones transferred from other inflectional classes, cf. e.g. the entire masculine singular paradigm of

22 The capital D designates the synchronic use of the form as a dative.

52
the Gothic i-stems where pivotal forms such as the nom.sg., acc.sg. and also the more marginal
voc.sg. have served as common points of reference for a total merger of the masculine singular of
the a-stem and i-stem paradigms or, in other words, for the introduction of the a-stem gen.sg. and
dat.sg. desinences in the i-stem paradigm. Whereas the intruding a-stem gen.sg. and dat.sg.
desinences have entirely replaced the original i-stem desinences in the masculine singular paradigm
of Gothic, they serve only as options in the masculine singular of the i-stem paradigms of the
remaining Germanic languages. Another example is the desinence of the nom.sg. of feminine i-
stems in Old Norse, i.e. originally ON -r < PG *-iz as in the corresponding masculine form, which
has been replaced by the feminine ō-stem nom.sg. desinence ON -Ø (u-umlauting) < PG *- in
most forms with the original desinence ON -r being retained in only a marginal group, cf. e.g.
Krahe (1967: 29) and Boutkan (1995: 241).
Bearing in mind the many points of reference between the Germanic i-stems and other
inflectional classes, above all the root nouns and the remaining vocalic stems, we can hardly be
surprised that numerous transitions of lexemes from the i-stems to these other inflectional classes
and vice versa have taken place in most if not all ancient Germanic languages, cf. e.g. Thöny (2013:
66-70) for a brief overview of the transitional tendencies operating in the i-stems. For the attempt of
historical and comparative linguists to separate archaisms from innovations, such working
conditions strongly complicate the task. Without any further clues than inflectional information,
historical and comparative linguists are faced with the insoluble task of determining whether a
given noun should be categorised and reconstructed as, say, a masculine a- or i-stem, a masculine a-
or ja-stem, or a feminine i- or ō-stem. Consequently, they must transcend the field of inflectional
morphology in their search for further clues that might help them determine the exact derivational
prehistory of a given noun.
As I intend to show in the present article, one area of linguistics that might provide the historical
and comparative linguist with additional information of this kind is the crossfield of derivational
morphology and phonotaxis. Hence follows that the purpose of the present article is to facilitate the
determination of the derivational character of Proto-Germanic nouns by setting a range of criteria
for the shape, structure and function of Proto-Germanic primary i-stems inherited from Proto-Indo-
European.

53
2. Form and function of the Germanic primary i-stems

2.1. Masculine i-stem verbal abstracts derivationally matching the ablaut grade found in the stem of
the preterite participle of a corresponding strong verb

Most grammars of Proto-Germanic and accordingly most scholars occupied with Germanic
derivational morphology have noticed that the bulk of Germanic i-stems appear in the masculine
gender with radical zero grade with voiced Verner’s variants root-finally as an indication that the
Proto-Indo-European accent was located on the i-stem suffix, and with the function of verbal
abstracts, i.e. action nouns which have often been secondarily concretised as resultative nouns, to
strong verbs, cf. e.g. Krahe & Meid (1967: 65-66), Hinderling (1967), Bammesberger (1990: 128),
Schaffner (2001: 421-435) and Casaretto (2004: 166-169). Furthermore, such i-stem verbal
abstracts are often parred with ga-prefigated neuter a-stems functioning as collective abstracts, cf.
e.g. Hinderling (1967: 42-102).
However, not all of these i-stems appear with radical zero grade. In a considerable number of
cases, we find radical o-grade (PIE *o > PG *a), and even full grade and lengthened grade are
attested. Synchronically speaking, part of the reason for this mixture of different radical ablaut
grades is to be found in the circumstance that different ablaut grades are also found in the preterite
participles of strong verbs. Only in class I-IV strong verbs is the preterite participle formed with
radical zero grade, cf. e.g. the class I strong verb PG *gripena- ~ *gripana- s.v. ‘caught’ (pret.ptc.),
the class II strong verb PG *budena- ~ *budana- s.v. ‘bidden’ (pret.ptc.), the class III strong verb
PG *bundena- ~ *bundana- s.v. ‘bound’ (pret.ptc.) and the class IV strong verb PG *numena- ~
*numana- s.v. ‘taken’ (pret.ptc.). In the class V strong verbs, on the contrary, the preterite participle
is formed with radical full grade, cf. e.g. PG *setena- ~ *setana- s.v. ‘sat’ (pret.ptc.), and in the
preterite participle of class VI strong verbs, we encounter a synchronic a-vowel representing either
PIE *o or *H, cf. e.g. PG *takena- ~ *takana- s.v. ‘taken’ (pret.ptc.). The reduplicated strong verbs
complicate the ablaut picture even further, especially considering the fact that their preterite
participles may be shaped with synchronic radical PG *a, * or * in the class I-VI non-ablauting
reduplicated strong verbs and with PG * in the class VII ablauting reduplicated strong verbs. For
details on the ablaut system of Germanic strong verbs, which lies outside the scope of this article,
cf. e.g. Krahe (1967: 100-107, 114-115).
The following pages contain a hopefully close to complete list of primary Germanic i-stem
verbal abstracts derivationally matching the ablaut grade found in the stem of the preterite participle
of a corresponding strong verb according to the class of the strong verb.

54
2.1.1. I-stem verbal abstracts corresponding to class I strong verbs

1 PG *biti- m. ‘bite prick’. Represented in OE bite m. ‘id.’ OFris. bite m. ‘splinter’ OS biti
m. ‘id.’ OHG biz m. ‘id.’. Radical zero grade also in e.g. PG *bita- n. ‘bite’ > ON bit n.
‘id.’ OHG (gi-)piz n. ‘rein bridle’ and PG *bitan- m. ‘bite bit’ > ON biti m. ‘bite tooth’
OE bita m. ‘bite bit’ OHG bizzo m. ‘bite bit lump’. Derivationally matching the class I
strong verb PG *bītana- s.v. ‘bite’ > ON bíta s.v. ‘id.’ OE bītan s.v. ‘id.’ OFris. bīta s.v.
‘id.’ OS bītan s.v. ‘id.’ OHG bīzan s.v. ‘id.’. To be reconstructed as PIE *bhid-í- to the
root PIE *bhe d- ‘slit split up’ cf. further Skt. bhinátti v. ‘splits breaks destroys cuts up’
(< PIE *bhi-né-d-), Av. ast .b d- adj. ‘bone-breaking, bone-damaging’ Gr. φεί ομαι v.
‘spare safe refrain from’ Lat. f nd v. ‘slit split up’ etc. Literature: Bammesberger (1990:
128), Bjorvand & Lindeman (2000: 75-76), Boutkan & Siebinga (2005: 50-51),
Holthausen (1974: 24), IEW (2005: 116-117), Kluge/Seebold (2002: 106), Kroonen (2013:
65), Lehmann (1986: 66), LIV (2001: 70-71), Lloyd et al. (1998: 143-145, 148), Orel
(2003: 45-46), Philippa et al. (2003: 312), Seebold (1970: 96-99), Sehrt (1966: 54), De
Vries (1962: 38).
2 PG *bliki- m. ‘appearance emergence’. Represented in OE blice m. ‘exposure of a bone’
OHG blic, plic, blich etc. m. ‘gloss; look’ (may also be inflected as a masculine a-stem).
Radical zero grade also in e.g. PG *blika- n. ‘gloss sheen gleam; metal leaf’ > ON blik n.
‘id.’ OS blek n. ‘metal leaf’ OHG blech n. ‘metal leaf; gloss sheen gleam’.
Derivationally matching the class I strong verb PG *blīkana- s.v. ‘shine’ > ON blíkja s.v.
‘gleam’ OE blīcan s.v. ‘shine light’ OFris. blīka s.v. ‘be visible’ OS blīkan s.v. ‘shine’
OHG -blī an s.v. ‘shine radiate’. To be reconstructed as PIE *bhl ĝ-í- to the root PIE
*bhle ĝ- ‘shine’ only found in Germanic and Balto-Slavic, cf. Lith. bl k t v. ‘become
pale’ OCS blьstat v. ‘sparkle shine’. There is no need to reconstruct a laryngeal for this
root as suggested by Boutkan & Siebinga (2005: 53) and, albeit indirectly, by IEW (2005:
156-157), for the long *-ī- of the Balto-Slavic forms is easily explicable as PIE *-i-
lengthened by Winter’s Law. Literature: Bammesberger (1990: 129), Boutkan & Siebinga
(2005: 53), Holthausen (1974: 27), IEW (2005: 156-157), Kluge/Seebold (2002: 131-132),
Kroonen (2013: 69), LIV (2001: 89), Lloyd et al. (1998: 182-184), Orel (2003: 49),
Philippa et al. (2003: 327), Seebold (1970: 118-119), Sehrt (1966: 56), De Vries (1962:
44).

55
3 PG *gridi- f. ‘step standing’. Represented in Goth. griþs m. ‘standing’ MHG grit m.
‘step’. Derivationally matching the class I strong verb PG *grīdana- s.v. ‘wander’ > MHG
grīten s.v. ‘spread the legs wide’. Probably to be reconstructed as PIE *ghridh-í- to the root
PIE *ghre dh- ‘wander step’ cf. also e.g. Lith. grìdyti v. ‘go wander’ OCS gr st v. ‘go
come’ (pres. gr dǫ). Literature: Casaretto (2004: 185), IEW (2005: 456-457), Kroonen
(2013: 189), Lehmann (1986: 161), LIV (2001: 203), Orel (2003: 142).
4 PG *hrini- m. ‘touch’. Represented in OE hrine m. ‘touch feeling’ OFris. hrene m.
‘(sense of) smell’. Radical zero grade also in e.g. PG *hrina- n. ‘ornament construction’ >
OE (ge-)hrin n. ‘id.’. Derivationally matching the class I strong verb PG * rīnana- s.v.
‘touch effect’ > ON hrína s.v. ‘effect influence’ OE rīnan s.v. ‘touch reach strike’ OS
rīnan s.v. ‘touch’ OHG rīnan, rīnan s.v. ‘touch get hold of’. Often connected to Lith.
kr n f. ‘cream’ Latv. krèims m. ‘id.’ krìet v. ‘skim cream off milk’ (i.e. < ‘touch
gently/lightly brush’) and reconstructed as PIE *kri-n-(H)-í-, i.e. as an i-stem to a zero
grade nasal present form of a root PIE *kre H - ‘brush touch’ but according to LIV
(2005: 368), the Germanic forms should be separated from the Baltic ones which continue
a root PIE *kreh1 - ‘sift divide’. Literature: Bammesberger (1990: 56, 129), Boutkan &
Siebinga (2005: 182), Holthausen (1974: 174), IEW (2005: 618), Kroonen (2013: 241,
247), LIV (2005: 366-368), Orel (2003: 187), Seebold (1970: 271), De Vries (1974: 256).
5 PG *hwini- m. ‘whistling whiz’. Represented in ON hvinr m. ‘id.’. Derivationally
matching the class I strong verb PG * wīnana- s.v. ‘whizz whistle’ > ON hvína s.v. ‘id.’
OE wīnan s.v. ‘id.’. Probably of onomatopoetic origin. Literature: Bammesberger (1990:
129), Holthausen (1974: 182), IEW (2005: 628), Kroonen (2013: 267), Seebold (1970:
280), De Vries (1962: 273).
6 PG *lidi- m. ‘going’ vel sim. Represented in OHG (ab-)lit m. ‘death’ (uz-)lit m. ‘error’.
Radical zero grade also in e.g. PG *lida- n. ‘company escort; vessel’ > ON lið n.
‘company escort army’ OE lid n. ‘ship’ OFris. lid n. ‘crowd troop company’ and PG
lidan- m. ‘follower companion’ > ON liði m. ‘id.’ OE lida m. ‘skipper master of ship’.
Derivationally matching the class I strong verb PG *līþana- s.v. ‘go’ > Goth. (ga-)leiþan
s.v. ‘go come travel’ ON líðа s.v. ‘go; pass by pass away die’ OE līt an s.v. ‘go sail;
disappear’ OFris. līt a s.v. ‘go’ OS līt an s.v. ‘go leave; perish’ OHG līdan s.v. ‘go
pass; suffer tolerate endure’. To be reconstructed as PIE *lit-í- to the root PIE *le t- ‘go
away’ cf. also e.g. Toch. B lita v. ‘went away’ Av. - r θ e t v. ’dies’ Gr. λοίτη f. ‘burial

56
funeral’. Literature: Bammesberger (1990: 129), Bjorvand & Lindeman (2000: 528-529),
Holthausen (1974: 201, 204), IEW (2005: 672), Kluge/Seebold (2002: 568), Kroonen
(2013: 340), Lehmann (1986: 6), LIV (2001: 410), Lloyd & Springer (1988: 26), Orel
(2003: 248-249), Philippa et al. (2007: 229-230), Seebold (1970: 328-330), Sehrt (1966:
335-336), De Vries (1962: 354).
7 PG *siki- m. ‘sigh’. Represented in OE sice m. ‘id.’. Derivationally matching the class I
strong verb PG *sīkana- s.v. ‘sigh; long yearn’ > OE sīcan s.v. ‘id.’. Further etymology
uncertain. Literature: Bammesberger (1990: 129), Holthausen (1974: 291), Seebold (1970:
391).
8 PG *skridi- ~ *skriþi- m. ‘slide step’. Represented in ON skriðr m. ‘sliding movement (of
a ship)’ OE skriþe m. ‘run course’ OHG skrit m. ‘step’. Derivationally matching the
class I strong verb PG *skrīþana- ~ *skrīdana- s.v. ‘stride’ > ON skríða s.v. ‘stride creep
crawl’ OE scrīt an s.v. ‘go wander; move creep’ OFris. skrīda s.v. ‘walk’ OS skrīþan,
skrīdan s.v. ‘glide stride pace go’ OHG (bi- skrītan s.v. ‘id.’. To be reconstructed as PIE
*skrit-í- to the root PIE *skre t- ‘stride move’ and possibly related to Lith. skrìsti v. ‘rush
run’ skraidùs adj. ‘fast’ but these forms continue PIE *-dh- rather than the *-t- found in
the precursor of the Germanic forms. Consequently, LIV (2001: 563) suggests that a more
proper cognate be Lith. skríesti v. ‘move around in circles’ (pres. skríe ) < PIE
*skr t-e-. As the only deverbal i-stem abstract in Germanic, PG *skridi- ~ *skriþi- m.
‘slide step’ is represented with both the unvoiced and the voiced Verner’s variant of PIE
*t. All other i-stems only display the voiced variant indicating that the PIE stress was
originally situated on the suffix. In the strong verb PG *skrīþana- ~ *skrīdana- s.v.
‘stride’ the opposite situation prevails: Where only the unvoiced variant is expected i.e.
PIE *skr t-e- > PG *skrīþana-, also the voiced variant is found. Literature: Bammesberger
(1990: 129), Boutkan & Siebinga (2005: 355), Holthausen (1974: 283), IEW (2005: 935-
938, esp. 937), Kluge/Seebold (2002: 826), Kroonen (2013: 448-449), LIV (2001: 563),
Lloyd et al. (1998: 114), Philippa et al. (2009: 118), Seebold (1970: 421-422), Sehrt (1966:
474-475), De Vries (1962: 503).
9 PG *slidi- m. ‘misstep slip; error’. Represented in OE slide m. ‘id.’. Radical zero grade
also in e.g. PG *slidan- m. ‘sled’ > ON sleði m. ‘id.’ OS slido m. ‘id.’ OHG slito m. ‘id.’.
Derivationally matching the class I strong verb PG *slīdana- s.v. ‘misstep slide slip; err’
> OE slīdan s.v. ‘slide slip; err’ MHG slīten s.v. ‘id.’. To be reconstructed as PIE

57
*h3slidh-í- to the root PIE *h3sle dh- ‘slide; err’ cf. also e.g. Skt. srédhati v. ‘slips down
slides; errs’ Gr. ὀλισθάνω v. ‘slide slip’ ὀλισθός m. ‘slipperiness’ (both < PIE *h3lisdh-
with metathesis of *-s-, cf. Rasmussen (1992: 349), slidùs adj. ‘slippery’ slýsti v. ‘slip
slide’ OCS sl dъ m. ‘track trail’. Literature: Bammesberger (1990: 129), Holthausen
(1974: 298), IEW (2005: 960-961), Kluge/Seebold (2002: 810), Kroonen (2013: 454), LIV
(2001: 307), Orel (2003: 351), Philippa et al. (2009: 173), Seebold (1970: 427-428), De
Vries (1962: 514).
10 PG *sliki- m. ‘furrow’. Represented in OHG slih m. ‘id.’. Derivationally matching the
class I strong verb PG *slīkana- s.v. ‘creep’ > OHG slī an s.v. ‘id.’. To be reconstructed
as PIE *sl ĝ-í- to the root PIE *sle ĝ- ‘slide; err’ cf. also e.g. OIr. (fo-)slig adj. ‘smeared
daubed coated’ OCS slьzъkъ adj. ‘slippery’. Literature: Bammesberger (1990: 129),
Bjorvand & Lindeman (2000: 814-815), IEW (2005: 663-664), Kluge/Seebold (2002:
808), LIV (2001: 566-567), Orel (2003: 351), Seebold (1970: 428-429).
11 PG *sliti- m. ‘break split’. Represented in OE slite m. ‘bit scratch break’ OHG sliz m.
‘split destruction’. Radical zero grade also in e.g. PG *slita- n. ‘split’ > ON slit n. ‘id.’
OE (lah-)slit n. ‘violation (of the law)’, OFris. slit n. ‘slit break split’ OHG (gi-)sliz n.
‘split splitup disruption’. Derivationally matching the class I strong verb PG *slītana- s.v.
‘tear up tear to pieces’ > ON slíta s.v. ‘scratch break destroy’ OE slītan s.v. ‘tear up tear
to pieces; torment’ OFris. slīta s.v. ‘scratch; remove’ OS slītan s.v. ‘fray split’ OHG
slīzan s.v. ‘tear up tear to pieces’. Possibly connected to Lith. skleĩst v. ‘spread’ (pres.
skle dž ) and thus to be reconstructed as PIE *s(k)lid-í- to the root PIE *s k le d- ‘cut’.
Literature: Bammesberger (1990: 129), Holthausen (1974: 299), IEW (2005: 923-927, esp.
926), Kluge/Seebold (2002: 808, 811), Orel (2003: 351), Seebold (1970: 430-431), Sehrt
(1966: 480), De Vries (1962: 516).
12 PG *snidi- m. ‘cut’. Represented in OE snide m. ‘cut’ OFris. snid m. ‘id.’ OHG snit m.
‘cut grain crop’. Radical zero grade also in e.g. PG *snida- n. ‘cut’ > ON snið n. ‘slice
piece’ OE (ge-)snid n. ‘killing slaughter’. Derivationally matching the class I strong verb
PG *snīþana- s.v. ‘cut’ > Goth. sneiþan s.v. ‘kill slaughter sacrifice’ ON sníða s.v. ‘cut
off’ OE snīt an s.v. ‘cut; kill’ OFris. snīt a s.v. ‘cut’ OS snīt an s.v. ‘id.’ OHG snīdan
s.v. ‘cut; reap harvest’. ncertain etymology though possibly to be reconstructed as PIE
*snit-í- to the root PIE *sne t- ‘cut’ cf. also e.g. Cz. sn t m. ‘branch’. Literature:
Bammesberger (1990: 129), Boutkan & Siebinga (2005: 360), Holthausen (1974: 304),

58
IEW (2005: 974), Kluge/Seebold (2002: 818), Kroonen (2013: 462), Lehmann (1986: 316-
1317), LIV (2001: 574), Orel (2003: 358), Philippa et al. (2009: 202-203), Seebold (1970:
443-444), Sehrt (1966: 481), De Vries (1962: 524).
13 PG *stigi- m. ‘ascent rising’. Represented in OE stige m. ‘id.’. Radical zero grade also in
e.g. PG *stiga- n. ‘step’ > ON stig n. ‘id.’; PG st g (n)- f. ‘ascent rising’ > ON (upp-)stiga
f. ‘id.’ OHG stega f. ‘staircase latter’; and PG *stigan- m. ‘staircase ladder’ > ON stigi
m. ‘id.’. Derivationally matching the class I strong verb PG *stīgana- s.v. ‘rise ascend’ >
Goth. steigan s.v. ‘id.’ ON stíga s.v. ‘step tread’ OE stīgan s.v. ‘go ascend mount’
OFris. stīga s.v. ‘ascend’ OS stīgan s.v. ‘id.’ OHG stīgan s.v. ‘id.’. To be reconstructed as
PIE *stigh-í- to the root PIE *ste gh- ‘cut’ cf. also e.g. Gr. στείχω v. ‘go march rise’
στοίχος m. ‘line row’ Alb. shteg m. ‘path foot-way’ OIr. tíagu v. ‘go’ Lith. steĩgt v.
‘organise’ (dialectally ‘hurry’). Literature: Bammesberger (1990: 129), Bjorvand &
Lindeman (2000: 858), Boutkan & Siebinga (2005: 377-378), Holthausen (1974: 322),
IEW (2005: 1017-1018), Kluge/Seebold (2002: 879), Lehmann (1986: 324-325), Kroonen
(2013: 479), LIV (2001: 593-594), Philippa et al. (2009: 284), Seebold (1970: 466-467),
Sehrt (1966: 507-508), De Vries (1962: 547).
14 PG *stiki- m. ‘prick stab’. Represented in Goth. stiks m. (only in in stika melis ‘in a
moment’) OE stice m. ‘prick stab’ OFris. steke m. ‘id.’ OS stiki m. ‘id.’ OHG stih m.
‘id.’. Derivationally matching a lost verb PG *stikana- s.v. ‘bite’ i.e. a verb of the type
Skt. tudáti v. ‘thrusts pushes’ with radical zero grade. This verb has given rise to a
reinterpretation of the a-umlauted form PG *stekana- s.v. ‘stick thrust’ > OFris. steka s.v.
‘id.’ OS stekan s.v. ‘id.’ OHG stehhan s.v. ‘id.’ as a regular class V strong verb to which
a competing i-stem noun with radical o-grade has also been formed, viz. PG *staki- m.
‘mark scar’. To be reconstructed as PIE *(s)tig-í- to the root PIE * s te g- ‘prick sharp’
cf. also e.g. Skt. téjate v. ‘is sharp’ tigmá- adj. ‘pointed’ Av. tigra- adj. ‘id.’ Gr. στίγμα f.
‘dot point’ Lat. (in- stīg v. ‘stimulate instigate’. Literature: Boutkan & Siebinga (2005:
371-372), Casaretto (2004: 176-177), Holthausen (1974: 321), IEW (2005: 1016-1017),
Kluge/Seebold (2002: 878, 883), Kroonen (2013: 476), Lehmann (1986: 325-326), LIV
(2001: 592-593), Orel (2003: 370), Seebold (1970: 467-468), Sehrt (1966: 505).
15 PG *stridi- m. ‘pace step’. Represented in OE stride m. ‘id.’ MHG strit m. ‘id.’.
Derivationally matching the class I strong verb PG *strīdana- s.v. ‘resist fight quarrel’ >
OSw. stríða s.v. ‘fight quarrel’ OE strīdan s.v. ‘stride; get by force’ (semantically

59
influenced by PG *skrīþana- ~ *skrīdana- s.v. ‘stride’, OFris. strīda s.v. ‘fight quarrel’
OHG strītan s.v. ‘id.’. Further etymology uncertain, but possibly related to e.g. Lith.
strainùs adj. ‘unruly recalcitrant’ and to be reconstructed either as PIE *stri-dhh1-í- to a
compound PIE *stre -dhh1- ‘stand upright’ cf. Kroonen (2013: 485) or as PIE *stri-tí-, i.e.
a ti-stem abstract PIE to the root PIE *stre , which was later resegmented as a primary i-
stem abstract noun *strit-í- to which a new strong verb PG *strīdana- was created by
means of analogy. Given the validity of either analysis, PG *stridi- has no further
relevance to our purpose. Literature: Bammesberger (1990: 129), Boutkan & Siebinga
(2005: 380-381), Holthausen (1974: 326), IEW (2005: 1022-1027, esp. 1026),
Kluge/Seebold (2002: 890), Kroonen (2013: 485), Orel (2003: 381-382), Philippa et al.
(2009: 303), Seebold (1970: 475-476), Sehrt (1966: 509), De Vries (1962: 553).
16 PG *striki- m. ‘dot line’. Represented in Goth. striks m. ‘streak; hook serif’ OHG strih
m. ‘dot line stroke’. Radical zero grade also in e.g. PG *strika- n. ‘strip’ > ON strik n.
‘id.’ and PG *strikan- m. ‘dot sign’ > OE strica m. ‘id.’. Derivationally matching the class
I strong verb PG *strīkana- s.v. ‘stroke rub move’ > ON strýkva, strýkja s.v. ‘id.’ (with
secondary v), OE strīkan s.v. ‘id.’ OFris. strīka s.v. ‘id.’ OHG strī an s.v. ‘id.’. To be
reconstructed as PIE *strig-í- to the root PIE *stre g- ‘stroke sweap’ cf. also e.g. Lat.
str ng v. ‘skim scratch’ (< PIE *stri-n-g-), striga f. ‘swath windrow’ OCS str t v. ‘cut
slip’ (pres. str gǫ). Literature: Bammesberger (1990: 129), Casaretto (2004: 177),
Holthausen (1974: 326), IEW (2005: 1028-1029), Kluge/Seebold (2002: 890), Kroonen
(2013: 485), Lehmann (1986: 328), LIV (2001: 603-604), Orel (2003: 381-382), Philippa
et al. (2009: 304), Seebold (1970: 476-477), De Vries (1962: 554).
17 PG *swik(w)i- m. ‘delay postponement’ > OE swice m. ‘id’. Radical zero grade also in
e.g. PG *swik(w)a- n. ‘treason fraud’ > ON svik n. ‘id.’ OE swic n. ‘deception illusion’
and PG *swik(w)an- m. ‘traitor’ > OE swica m. ‘id.’. Derivationally matching the class I
strong verb PG *swīk w ana- s.v. ‘betray deceive’ > ON svíkva, svíkja s.v. ‘id.’ OE
swīcan s.v. ‘move about wander leave desert; betray deceive’ OFris. swīka s.v.
‘remove’ OS swīkan s.v. ‘desert leave in the lurch’ OHG swī an s.v. ‘turn away
decline’. The quality of the root final velar is uncertain: ON svíkva, svíkja points to a
labiovelar PG *kw while the remaining forms point to a plain velar PG *k. Bjorvand &
Lindeman (2000: 888) reasonably holds that the plain velar is original and that ON svíkva,
svíkja is secondarily transformed from *svíka by analogy starting in the pres. 3.sg. svíkr,

60
cf. the pattern of syngva ‘sing’ : syngr, blíkja ‘shine’ : blíkr and X : svíkr where X = blíkva.
Further etymology uncertain, but possibly to be reconstructed as PIE *s g-í- to the root
PIE *s e g- ‘give up abandon hesitate’ cf. also e.g. Lith. svaĩgt v. ‘become dizzy’ Ru.
sv gát’ v. ‘drift about’. Literature: Bammesberger (1990: 129), Bjorvand & Lindeman
(2000: 888-889), Boutkan & Siebinga (2005: 386), Holthausen (1974: 336), IEW (2005:
1042), LIV (2005: 608), Orel (2003: 396-397), Seebold (1970: 486-487), Sehrt (1966: 520-
521), De Vries (1962: 569).
18 PG *wliti- m. ‘face appearance form’. Represented in Goth. wlits m. ‘id.’ OE wlite m.
‘face appearance’ OFris. wlite m. ‘id.’ OS wliti ‘gloss light appearance form’; PG
*wlitu- m. > ON litr m. ‘appearance colour’; and PG *wl t - f. ‘visible part of the body’ >
OE wlitu f. ‘form visible wound’. Radical zero grade also in e.g. PG *wlita- n.
‘appearance’ > ON (á-)lit n. ‘id.’. Derivationally matching the class I strong verb PG
*wlītana- s.v. ‘see look’ > ON líta s.v. ‘look watch’ OE wlītan s.v. ‘see look’. To be
reconstructed as PIE * l d-í- to a root PIE * le d- with no further cognates unless
comparison is made to the simple root PIE * el- ‘see’. The element *-id- remains
unexplained, though, unless we choose to accept Hamp’s (1982: 79-81) analysis of PIE
* le d- as a contamination of PIE * el- ‘see’ and * e d- ‘id.’. Literature: Boutkan &
Siebinga (2005: 455), Casaretto (2004: 178), IEW (2005: 1136-1137), Holthausen (1974:
403), Kluge/Seebold (2002: 50-51), Kroonen (2013: 591), Lehmann (1986: 408), Lloyd &
Springer (1988: 280-283), Orel (2003: 469), Seebold (1970: 563-564), Sehrt (1966: 712),
De Vries (1962: 358-359).
19 PG *writi- m. ‘sign letter’. Represented in Goth. writs m. ‘id.’ OHG riz m. ‘stroke letter’
(hapax, only attested in acc.sg. ainana writ ‘one sign’). Radical zero grade also in e.g. PG
*writa- n. ‘writing book’ > ON rit n. ‘writing’ OE writ n. ‘writing book’. Derivationally
matching the class I strong verb PG *wrītana- s.v. ‘carve inscribe draw write’ > ON ríta
s.v. ‘id.’ OE wrītan s.v. ‘id.’ OFris. wrīta s.v. ‘id.’ OS wrītan s.v. ‘id.’ OHG rīzan s.v.
‘tear write’. As for the further etymology, Frisk (1944: 28-30) has suggested that PG
*wrītana- should be compared to Arm. ergicanem v. ‘bite tear’. Alternatively Kroonen
(2013: 597) regards the strong verb PG *wrītana- as secondarily formed from the iterative
verb PG *wr tt ja na- w.v. II ‘carve write’ (< Pre-PG *writ-n-) which is in turn derived
from PG *wrīþana- s.v. ‘twist turn’ i.e. < PIE * re t- ‘twist turn’ also found in e.g. Lith.
r ẽst v. ‘bend curve’ (pres. r e ). Literature: Bammesberger (1990: 130), Casaretto

61
(2004: 178), Holthausen (1974: 409), IEW (2005: 1159-1160, 1163-1164), Kluge/Seebold
(2002: 755), Kroonen (2013: 597), Lehmann (1986: 410-411), LIV (2001: 700), Orel
(2003: 472-473), Seebold (1970: 566-567), Sehrt (1966: 721-722), De Vries (1962: 448).

2.1.2. I-stem verbal abstracts corresponding to class II strong verbs

20 PG *druzi- m. ‘fall’. Represented in Goth. drus m. ‘id.’ OE dryre m. ‘fall decline’. As

remarked by Schaffner (2001: 428-429), the existence of Goth. drus with voiceless s by no
means guarantees an original paradigm PG *drusi- ~ *druzi- continuing a PIE paradigm
with mobile accent. The unvoiced Verner’s variant in Gothic may simply result from
Auslautsverhärtung or have spread from the strong verb to the i-stem abstract; a
development paralleled by primary deverbative derivations such as abstract nouns,
causatives and verbal adjectives, cf. e.g. the causative Goth. gadrausjan w.v. I ‘plunge
down’ vs. MHG tr ren w.v. I ‘drip ooze shed’ < PG *drauzjana- w.v. I ‘make fall’.
Radical zero grade also in e.g. PG *druza- n. ‘glaucoma cataract’ > ON drør n. ‘id.’ which
might be compared to PG *druzi-. Derivationally matching the class II strong verb PG
*dreusana- s.v. ‘fall’ > Goth. driusan s.v. ‘fall fall down’ OE dr osan s.v. ‘rush fall
perish’ OS driosan s.v. ‘fall’. If related to the stative PG *drūs (ja)na- w.v. III ‘fall
asleep’ > OE drūs an w.v. III ‘become slow’ and thus further to Ru. drýxn t’ v. ‘sleep’ as
posited by Kroonen (2013: 105), the root should be reconstructed as PIE *dhre Hs- ~
*dhruHs- ‘fall asleep sleep’(?) and PG *druzi- would have to be regarded as secondarily
formed from the weak stem, i.e. the stem of the preterite participle, of the strong verb PG
*dreusana-. Literature: Bammesberger (1990: 130), Casaretto (2004: 174), Holthausen
(1974: 77-79), IEW (2005: 274-275), Kroonen (2013: 105), Lehmann (1986: 95), LIV
(2001: 157-158), Orel (2003: 76-77), Philippa et al. (2005: 195), Seebold (1970: 170-171),
Sehrt (1966: 84), De Vries (1962: 86).
21 PG *flugi- m. ‘escape’. Represented in Goth. þlauhs m. ‘id.’ ON flugr m. ‘id.’ OE flyge
m. ‘id.’ OS flugi m. ‘id.’. Radical zero grade also in e.g. PG *flugan- m. ‘escapee’ > ON
(fuð-)flogi m. ‘runaway bridegroom’ and parallelly (flann-)fluga f. ‘runaway bride
runaway wife’ (< PG *fl g n-). Derivationally matching the class II strong verb PG
*fleuhana- s.v. ‘flee escape’ > Goth. þliuhan s.v. ‘id.’ OE fl on s.v. ‘id.’ OFris. fl s.v.
‘flee disappear’ OS fliohan s.v. ‘flee’ OHG fliohan s.v. ‘flee avoid’. The question
whether the cluster PG *fl- or *þl- is original has caused much debate. In my view, Matzel
(1962: 222-224, 237) has presented the most convincing solution to the problem, viz. that

62
PG *fl- > Goth. *þl- when followed by a vowel and a spirantic element. If so, PG *flugi- is
to be reconstructed as PIE *pluk-í- to the root PIE * le k- ‘hover be carried somewhere
through air or water; move (swiftly)’ cf. e.g. Lith. laũkt v. ‘swim move through the air’.
Literature: Bammesberger (1990: 130), Bjorvand & Lindeman (2000: 238-239), Boutkan
& Siebinga (2005: 120-121), Casaretto (2004: 177), Holthausen (1974: 107, 110), IEW
(2005: 837), Kluge/Seebold (2002: 301), Kroonen (2013: 544), Lehmann (1986: 363-364),
LIV (2001: 488), Lloyd & Lühr (2007: 389-392), Orel (2003: 107), Philippa et al. (2009:
539, 546-547), Seebold (1970: 517-518), Sehrt (1966: 136), De Vries (1962: 134).
22 PG *gruzi- m. ‘fear dread’. Represented in OE gryre m. ‘id.’ OS gruri m. ‘fear dread;
violence’. According to Bammesberger (1990: 130) derivationally matching the class II
strong verb PG *greusana- s.v. ‘frighten storm; overwhelm(?)’ > OE (be)-groren s.v.
‘overwhelmed’ (pret.ptc.). Further etymological connection uncertain. Holthausen (1974:
138) compares OE (be-)groren and gryre with Gr. ἔχαρον v. ‘attacked charged’ Lat. in-
gr v. ‘rush on storm’ and Lith. griáuti v. ‘tear down destroy’ derived from the root
reconstructed by LIV (2001: 202) as PIE *ghreh1 - ‘rush on’. IEW (2005: 435-454) prefers
a comparison of at least the Germanic i-stem forms with Skt. ghorá- adj. ‘awful terrible
dreadful’ n. ‘horror terror; magic spell’ and Goth. gaurs adj. ‘sad’; both from PIE *gho -
ró- ‘dreadful filled with fear’ which though a semantic near to perfect match is no
formally well-fitting candidate for a precursor to OE gryre and its closest Germanic
relatives. In the light of the obvious semantic affinities between ‘dreadful filled with fear’
and ‘rush on’ > ‘frighten storm; overwhelm’ I would not be surprised if the root
underlying the Germanic lexemes in question had arisen as a contamination of the two
forms PIE *ghrh1u- > *ghruh1- (zero grade of PIE *ghreh1 -) and PIE *gho -ró- and were
to be reconstructed as PG *greur- rather than the PG *greus- suggested by Bammesberger
(1990: 130).23 Literature: Bammesberger (1990: 130), Holthausen (1974: 138-139), IEW
(2005: 453-454), Sehrt (1966: 212).

23 There are no Germanic lexemes or inflected forms in unequivocal support of Bammesberger’s notion that the root-
auslauting consonant of NWG *greur- is a voiced Verner’s variant (NWG r < PG *z) of PG *s rather than a
continuant of plain PG *r. The closest candidate with comparable semantics is PG *geusana- s.v. ‘tremble’ > ON
gjósa s.v. ‘id.’ cf. e.g. De Vries (1962: 170-171) and Bjorvand & Lindeman (2000: 329), but its lack of a radical r
renders any attempt to compare these two forms futile.

63
23 PG *guti- m. ‘filling pouring’. Represented in OE gyte m. ‘id.’ OFris. gete m. ‘id.’ OHG
guz m. ‘fusiō i.e. outpouring’. Radical zero grade also in e.g. PG *guta- m. ‘pouring’ >
OHG (uz-)koz m. ‘id.’; PG *gutan- m. ‘pourer’ > OE lead-gota m. ‘lead-pourer’; and
further probably PG *guta(n)- m. ‘Goth’ > Goth. gut-þiuda f. ‘Gothic people’ ON goti m.
‘Goth’ OSw. gutar, gotar m. ‘people’ (pl.) OE gotan ‘Goths’ (pl.) cf. e.g. Strid (2008:
25-29) for a recent treatment of that issue. Derivationally matching the class II strong verb
PG *geutana- s.v. ‘pour’ > Goth. giutan s.v. ‘id.’ ON gjóta s.v. ‘drop throw, cast young’,
OE g otan s.v. ‘pour pour out’ OFris. -j ta s.v. ‘id.’ OS giotan s.v. ‘pour spill’, OHG
giozan s.v. ‘pour’. To be reconstructed as PIE *ĝhud-í- to the root PIE *ĝhe -d- ‘pour’
attested with or without the dental root extension in e.g. Skt. juhóti v. ‘pours sacrifices’
Toch. AB ku- v. ‘pour offer a libation’ Gr. χέω v. ‘pour shed; smelt’ and Lat. f nd v.
‘pour’. Literature: Bammesberger (1990: 130), Bjorvand & Lindeman (2000: 329-330),
Holthausen (1974: 128, 135, 141), IEW (2005: 448), Kluge/Seebold (2002: 357, 379),
Kroonen (2013: 177, 196), Lehmann (1986: 163-165), LIV (2001: 179-180), Orel (2003:
147), Seebold (1970: 228-229), Sehrt (1966: 188), De Vries (1962: 171, 183).
24 PG *hluti- m. ‘share lot’. Represented in ON hlutr m. ‘id.’ OE hlyte m. ‘id.’ OHG hluz,
luz m. ‘id.’. Radical zero grade also in e.g. PG *hluta- n. ‘share lot’ > OE hlot n. ‘id.’
OFris. hlot n. ‘id.’ OHG hloz, loz n. ‘id.’. Derivationally matching the class II strong verb
PG *hleutana- s.v. ‘obtain by lot’ > ON hljóta s.v. ‘get obtain’ OE l otan s.v. ‘cast lots;
get obtain by lot’ OS hliotan s.v. ‘acquire receive’ OHG hliozan, liozan s.v. ‘draw lots’.
A competing i-stem noun with radical o-grade also exists, viz. PG *hlauti- m. ‘share lot’.
Further etymology uncertain, but maybe to be analysed as PIE *k̂l H-d(h3)-í- to a
compound PIE *k̂le H-d(h3)- ‘give by lot’ consisting of the roots PIE *k̂le H-, cf. e.g.
Lith. kl ūt v. ‘stick’ (< PIE *kluH-) and PIE *deh3- ‘give’ cf. e.g. Skt. dád t v. ‘gives’
Gr. ί ωμι v. ‘give’ Lat. d v. ‘id.’ Lith. dúoti v. ‘id.’. Given this etymology PG *hluti-
must have received its short *u in analogy with the general ablaut pattern of class II strong
verbs, i.e. PG *hleutana- : X ~ *geutana- : *guti- with X equalling *hluti-. Literature:
Bammesberger (1990: 133), Bjorvand & Lindeman (2000: 543-544), Casaretto (2004:
180), Holthausen (1974: 162, 164-165), IEW (2005: 604-605, esp. 605), Kluge/Seebold
(2002: 582), Kroonen (2013: 230, 233), LIV (2001: 105-106, 365), Orel (2003: 176, 178),
Philippa et al. (2007: 263-264), Seebold (1970: 264-265), Sehrt (1966: 262), De Vries
(1962: 238, 240).

64
25 PG *hruzi- m. ‘fall death’. Represented in OE hryre m. ‘id.’. Radical zero grade also in
e.g. PG *hruza- n. ‘dead body corpse’ > ON hrør n. ‘id.’. Derivationally matching the
class II strong verb PG *hreusana- s.v. ‘fall’ > OE r osan s.v. ‘id.’ OHG hriusu s.v.
‘reor’ (leg. ‘ruor’ i.e. ‘fall’). To be reconstructed as PIE *krus-í- to the extended root PIE
*kre -s- ‘strike thrust’ cf. e.g. Gr. κρούω v. ‘strike smite’ Lith. kr t v. ‘trample smash
crush’ OCS sъ- kr t v. ‘destroy’ and without the extension PG *hruna- n. ‘fall’ > Icel.
hrun n. ‘id.’ and PG *hrunjana- w.v. I ‘fall’ > ON hrynja w.v. I ‘id.’. Literature:
Bammesberger (1990: 130), Holthausen (1974: 173, 176-177), IEW (2005: 622-623), LIV
(2001: 371), Orel (2003: 186), Seebold (1970: 276-277), De Vries (1962: 263).
26 PG *kuzi- m. ‘choice’. Represented in OE cyre m. ‘selection choice free will’ OFris. kere
m. ‘choice regulations’ OS (self-)kuri m. ‘arbitrariness’ OHG kuri f. ‘selection, choice,
consideration’. Radical zero grade also in e.g. PG *kuza- n. ‘choice election’ > ON kør n.
‘id.’. Derivationally matching the class II strong verb PG *keusana- s.v. ‘test trial; select,
prefer’ > Goth. kiusan s.v. ‘prove by trial test’ ON kjósa s.v. ‘choose select elect’ OE
c osan s.v. ‘choose elect’ OFris. k sa s.v. ‘choose establish’ OS kiosan s.v. ‘choose
recognise’ OHG kiosan s.v. ‘choose examine judge’. To be reconstructed as PIE *ĝ s-í-
to the root PIE *ĝe s- ‘taste’ also found in e.g. Hitt. k k z v. ‘tastes’ Skt. j áte v.
‘enjoys likes’ Gr. γεύομαι v. ‘tastes’ Lat. gustus m. ‘taste’. Literature: Bammesberger
(1990: 130), Boutkan & Siebinga (2005: 213, 215), Holthausen (1974: 47, 68), IEW (2005:
399-400), Kluge/Seebold (2002: 487, 548), Kroonen (2013: 286), Lehmann (1986: 219),
LIV (2001: 166-167), Philippa et al. (2007: 57), Seebold (1970: 293-294), Sehrt (1966:
304-305), De Vries (1962: 312, 341).
27 PG *lugi- m. ‘lie deception’. Represented in OE lyge m. ‘id.’ OHG lug m. ‘id.’.
Derivationally matching the class II strong verb PG *leugana- s.v. ‘lie’ > Goth. liugan s.v.
‘id.’ ON ljúga s.v. ‘id.’ OE l ogan s.v. ‘id.’ OFris l ga s.v. ‘id.’ OS liogan s.v. ‘id.’
OHG liogan s.v. ‘id.’. To be reconstructed as PIE *lugh-í- to the root PIE *le gh- ‘lie’ cf.
also e.g. OCS lъgat v. ‘lie’ lъžь adj. ‘lying false mendacious’ m. ‘liar’ lъža f. ‘lie’.
Literature: Bammesberger (1990: 130), Bjorvand & Lindeman (2000: 561), Boutkan &
Siebinga (2005: 239), Holthausen (1974: 199, 208), IEW (2005: 686-687), Kluge/Seebold
(2002: 584), Kroonen (2013: 333), Lehmann (1986: 235-236), LIV (2001: 417), Orel
(2003: 242), Philippa et al. (2007: 226), Seebold (1970: 336-337), Sehrt (1966: 341, 342),
De Vries (1962: 361).

65
28 PG *luzi- m. ‘loss’. Represented in OE lyre m. ‘id.’. Radical zero grade also in e.g. PG
*lusa- n. ‘loss’ > ON los n. ‘dissolution’ OE los n. ‘loss’ OHG (far-)los n. ‘id.’ and PG
*luza- m. ‘ruin destruction’ > OE (for-)lor m. ‘id.’ OS (far-)lor m. ‘id.’ OHG (far-)lor m.
‘id.’. Derivationally matching the class II strong verb PG *leusana- s.v. ‘lose; loosen’ >
Goth. (fra-)liusan s.v. ‘id.’ OE (for-)l osan s.v. ‘id.’ OFris. (for- l sa s.v. ‘id.’ OS (far-)
liosan s.v. ‘id.’ OHG (fir-)liosan s.v. ‘id.’. To be reconstructed as PIE *lu-s-í- to the
extended root PIE *le H-s- ‘set free let loose loosen’ found without the extension in e.g.
Hitt. lūr - c./n. ‘loss of honour disgrace humiliation’ Skt. l n t v. ‘cuts cuts off’ Gr.
λύω v. ‘loosen liberate set free let loose’ Lat. l v. ‘suffer make amends for’. The
laryngeal is obviously missing in the Germanic zero grade form PG *luzi- < PIE *lu-s-í-; a
PIE *luH-s-i- would have yielded PG †lūz -. The short radical *u has probably been
introduced from the stem of the preterite participle of PG *leusana-, i.e. PG *luzana-,
whose *u is in itself secondarily shortened from PG *ū in order for PG *luzana- to adapt to
the standard ablaut pattern eu : au : u : u of the class II strong verbs. Literature:
Bammesberger (1990: 130), Boutkan & Siebinga (2005: 227), Holthausen (1974: 199-200,
209), IEW (2005: 681-682), Kluge/Seebold (2002: 954), Kroonen (2013: 334), Lehmann
(1986: 123-124), LIV (2001: 417), Orel (2003: 243, 251), Philippa et al. (2007: 262-263),
Philippa et al. (2009: 503), Seebold (1970: 339-340), Sehrt (1966: 344-345), De Vries
(1962: 366-367).
29 PG *ruki- m. ‘smell; smoke’. Not attested in any of the ancient Germanic languages but
only later in e.g. MLG roke m. ‘smell’ MHG ruch m. ‘id.’. Seebold (1970: 380) however
mentions one older form, viz. OHG rug m. ‘smoke’ which does not seem to find support
in the scholarly literature. Derivationally matching the class II strong verb PG *reukana-
s.v. ‘smoke’ (originally ‘smell’) > ON rjúka s.v. ‘smoke smell; be driven be flung fly’
OE r ocan s.v. ‘reek; smoke steam’ OFris. r ka s.v. ‘smoke’ OHG riochan, riohhan s.v.
‘smoke steam’ but its ablaut grade is aberrant. A competing i-stem noun with radical o-
grade also exists, viz. PG *rauki- m. ‘smoke’. Further etymology uncertain, but the Baltic
languages offer some possible comparanda, e.g. Lith. ráugti v. ‘sour make sour’ and Latv.
rûgt v. ‘sour make sour; smoke’ where however the meaning ‘smoke’ is probably caused
by influence from another nearly homophonous verb represented by Lith. rũkt v. ‘smoke’.
Literature: Bammesberger (1990: 133), Bjorvand & Lindeman (2000: 742-743),
Holthausen (1974: 257), IEW (2005: 871-872), Kluge/Seebold (2002: 350, 764-765),

66
Kroonen (2013: 410), Orel (2003: 303, 308), Philippa et al. (2007: 659, 689-690), Seebold
(1970: 379-380), De Vries (1962: 449).
30 PG *skuti- m. ‘shot shooting’. Represented in ON skutr m. ‘stern’ (< ‘pushing forward’)
OE scyte m. ‘shot shooting’ OFris. skete m. ‘id.’ OHG scuz m. ‘shot speed’. Radical
zero grade also in e.g. PG *skuta- n. ‘throw shot’ > ON skot n. ‘shot weapon’ OFris. scot
n. ‘throw bullet’ OHG (ge-)scoz n. ‘bullet’. Derivationally matching the class II strong
verb PG *skeutana- s.v. ‘shoot’ > ON skjóta s.v. ‘shoot throw; pay’ OE sc otan s.v.
‘shoot throw; run’ OFris. sk ta s.v. ‘shoot supply’ OHG skiozan s.v. ‘shoot hurl
throw’. Further etymological connections remain disputed. Two often mentioned
possibilities are to either compare PG *skuti- to Skt. códati v. ‘pushes’ and Alb. hedh,
hjedh v. ‘throws shoots’ (< PAlb *ske d- and *sked-, respectively) and reconstruct it as
PIE *skud-í- to the root PIE *ske d- ‘throw shoot’ or to compare it to Lith. á t , á t v.
‘shoot run fast’ (< PIE *k̂e H-). Kroonen (2013: 445) alternatively suggests that PG
*skeutana- is actually formed on the basis of the iterative PG *sk tt ja na- (< PG *skuT-
n-), in itself a secondary zero grade formation from the root PIE *sket-. Literature:
Bammesberger (1990: 130), Bjorvand & Lindeman (2000: 810-811), Boutkan & Siebinga
(2005: 350), Holthausen (1974: 276, 286), IEW (2005: 955-956), Kluge/Seebold (2002:
802, 824), Kroonen (2013: 445, 451), LIV (2001: 560), Orel (2003: 339), Philippa et al.
(2009: 88, 90), Seebold (1970: 417-418), De Vries (1962: 496, 500, 508).
31 PG *slupi- m. ‘slip; piece of garment to slip over one’s head’ > OE (ofer-)slype m.
‘chasuble’. Radical zero grade also in e.g. PG *slupa- m./n. ‘slip; piece of garment to slip
over one’s head’ > ON sloppr m. ‘loose gown’ OE (ofer-)slop n. ‘id.’. Derivationally
matching the class II strong verb PG *sleupana- ~ *slū ana- s.v. ‘sneak slip’ > Goth.
sliupan s.v. ‘slip slip into slide’ OE slū an s.v. ‘id.’ OHG sliofan s.v. ‘slip sneak’.
Often compared to Lat. lūbr c s adj. ‘slippery’ but Kroonen (2013: 454) presents the
alternative view that it should rather be compared to Lith. į-slupti v. ‘slips into’ by means
of a reconstruction PIE *slup- to the root PIE *sle - ‘slip’ and with the comparison of
Lith p to PG *p explained by Kluge’s Law i.e. by the unattested form PG *sl ja na-
(< PG *sluP-n-) indirectly attested in the jan-verb PG *slup(p)iana- > OHG (int-)slupfen
w.v. I ‘get away’. Literature: Bammesberger (1990: 130), Holthausen (1974: 300), IEW
(2005: 963-964), Kluge/Seebold (2002: 810, 812), Kroonen (2013: 454), Lehmann (1986:
315), LIV (2001: 567), Orel (2003: 350), Seebold (1970: 435-436), De Vries (1962: 517).

67
32 PG *sudi- m. ‘decoction extract’. Represented in OE syde m. ‘id.’ OHG (salz-)suti m.
‘salina saltworks’. Radical zero grade also in e.g. PG *suþa- n. ‘broth soup’ > ON soð n.
‘meat broth’, OE (ge-)sod n. ‘cooking boiling dish’ (< PG *suda-), OFris. soth n. ‘broth’
OHG (ki-)sod n. ‘dish’. Derivationally matching the class II strong verb PG *seuþana- s.v.
‘boil seethe’ > ON sjóða s.v. ‘cook boil’ OE s ot an s.v. ‘boil seethe’ OFris. s t a s.v.
‘id.’ OHG siodan s.v. ‘id.’. A competing i-stem noun with radical o-grade also exists, viz.
PG *saudi- m. ‘meat broth’. To be reconstructed as PIE *h2sut-í- to the root PIE *h2se t-
‘boil bubble; move vigorously’ cf. also e.g. Gr. ἀ τμή f. ‘steam’ (< PIE *h2sut-méh2-),
OLith. siausti v. ‘storm rumble roar’ Lith. s aũst v. ‘play rage’ (pres. s a ), Ru. tít’
v. ‘play joke around’. Literature: Bammesberger (1990: 130), Bjorvand & Lindeman
(2000: 896-897), Holthausen (1974: 290, 306, 339), IEW (2005: 914-915), Kluge/Seebold
(2002: 847), Kroonen (2013: 435), LIV (2001: 285), Orel (2003: 326, 388), Philippa et al.
(2009: 663), Seebold (1970: 400-401), De Vries (1962: 478, 528).
33 PG *supi- m. ‘drink’. Represented in OE sype m. ‘id.’. Radical zero grade also in e.g. PG
*supan- m. ‘swallow gulp’ > ON sopi m. ‘swallow gulp mouthful’ OE sopa m. ‘sup
draught’. Derivationally matching the class II strong verb PG *sū ana- s.v. ‘swallow
quaff guzzle’ > ON súpa s.v. ‘sip drink’ OE sū an s.v. ‘swallow quaff guzzle slurp’
OHG sūfan s.v. ‘swallow quaff guzzle slurp drink’. Comparable to Skt. sū a- m. ‘broth
soup’ < PIE *suHp- to the root PIE *se H - which might be analysed as PIE *se H-
‘pour’ as seen in Hitt. a v. ‘pours’ compounded with PIE *p(h3)- ‘drink’ as seen in
e.g. Skt. píbati v. ‘drinks’ Lat. b b v. ‘drink’ (both < PIE *pi-ph3-é-), cf. Kroonen (2013:
493); given that analysis, however, the p of Skt. sū a- must have been reintroduced from
the general verbal root Skt. - since PIE *-ph3- > PIE *-b-, cf. again PIE *pi-ph3-é- >
*pi-be- in Skt. píbati, Lat. b b . The comparison of Skt. p to PG *p can thus be regular, but
it can also be explained through influence from the geminate arisen by Kluge’s Law in the
iterative PG *s ja na- ~ *s b ja na- w.v. II ‘soak’ (< PG *suP-n-). Kroonen further
explains the short *u of PG *s ja na- as a result of dissimilatory loss of the first
laryngeal of the form PIE *suH-ph3-. Another solution is, in my view, that the long *ū of
PG *sū ana- is actually to be regarded not as secondary but as regularly continuing PIE
*suH-ph3-é-, i.e. a present of the Skt. tudati-type, which was reanalysed in Proto-Germanic
as belonging to the class II strong verbs on the basis of examples such as PG *lūkana- s.v.
‘close’. Given the validity of this analysis the i-stem PG *supi- must be regarded as

68
secondary. Literature: Bammesberger (1990: 131), Bjorvand & Lindeman (2000: 878),
Holthausen (1974: 307, 330, 340), IEW (2005: 913), Kluge/Seebold (2002: 787), Kroonen
(2013: 493), Orel (2003: 388), Philippa et al. (2009: 220, 677), Seebold (1970: 399-400),
De Vries (1962: 530, 562).
34 PG tugi- m. ‘pull draw’. Represented in OE (on-)tyge m. ‘prohibited action’ (of-)tige m.
‘removal deprivation’ OHG zug m. ‘pull draw lead’. Radical zero grade also in e.g. PG
*tuga- n. ‘line rope rein bridle’ > ON tog n. ‘id.’ OE (ge-)tog n. ‘pulling’ and PG
*tugan- m. ‘puller leader’ > ON (her-)toge m. ‘duke’ OE (here-)toga m. ‘id.’ (folc-)toga
m. ‘leader (of the people)’ OFris. (her-)toga m. ‘duke’ OS (heri-)togo m. ‘id.’ OHG
(heri-)zogo, (heri-)zoho m. ‘id.’. Derivationally matching the class II strong verb PG
*teuhana- s.v. ‘lead pull draw’ > Goth. tiuhan s.v. ‘lead bring’ ON toginn s.v. ‘drawn’
(pret.ptc.), OE t on s.v. ‘pull draw’ OFris. tia s.v. ‘pull draw; bear beget’ OS tiohan s.v.
‘pull draw; bring up’ OHG ziohan s.v. ‘pull draw; provide for keep foster’. To be
reconstructed as PIE *duk-í- to the root PIE *de k- ‘lead’ cf. also e.g. Alb. nduk v. ‘pulls
out plucks out tears (esp. of hair)’ Lat. dūc v. ‘lead bring’ dux m./f. ‘leader’ MW dwc
v. ‘brings’. Literature: Bammesberger (1990: 131), Boutkan & Siebinga (2005: 407),
Holthausen (1974: 345, 350), IEW (2005: 220-221), Kluge/Seebold (2002: 409, 1011,
1018), Kroonen (2013: 515), Lehmann (1986: 346), LIV (2001: 124), Orel (2003: 405),
Philippa et al. (2005: 424-425), Philippa et al. (2009: 385), Seebold (1970: 503-505), Sehrt
(1966: 252, 536-537), De Vries (1962: 594).

2.1.3. I-stem verbal abstracts corresponding to class III strong verbs

35 PG *brungi- m. ‘bringing’. Represented in OS -)brung m. ‘return’ OHG (heim-)

prunc m. ‘income’. Derivationally matching the class III strong verb PG *bringana- s.v.
‘bring’ > Goth. briggan s.v. ‘id.’ OE bringan s.v. ‘id.’ OFris. bringa s.v. ‘id.’ OS
bringan s.v. ‘id.’ OHG bringan s.v. ‘id.’. Suggested by Brugmann (1901: 154-158) with
later modification by García Ramón (1999: 65-67) to be a combination of the root PIE
*bher- ‘carry’ and PIE *enek̂-, *enk̂-, *nek̂- (i.e. *h2nenk̂-) ‘reach’ cf. the parallel mixture
of W (he-)brwng v. ‘bring lead’ and the suppletion of Gr. φέρω v. ‘carry’ (present stem <
PIE *bher- ‘carry’) and Gr. ἤνεγκα, ἤνεγκον v. ‘carried’ (aorist stem < Pre-Gr. *enenk- <
*anenk- < PIE *h2nenk̂- ‘reach’). A peculiarity of PG *bringana- is that its preterite and
preterite participle stems are created with the dental suffix of the weak verbs, i.e. PG
*branh-t- > *br -t-. Strong preterite and preterite participle forms also occur, but they are

69
limited to OHG brang, brungum, brungan and OE brungen. García Ramón (1999: 67) sets
up a plausible scenario of the development that might have led to the making of the
attested forms. He assumes that PG *berana- s.v. ‘carry’ once co-existed with a causative
PG *angjana- w.v. I ‘let reach; bring’ (< PIE *h2onk̂- e-) whose preterite stem would
have been PG *anh-t- > * -t- vel sim. These two verbs coalesced in PG *brangiana- w.v.
I ‘bring’ with a pret. PG *branh-t- > *br -t-. In analogy with the ablaut in PG *berana-
(pret. bar), a new e-grade present stem PG *brengana- > *bringana- was created that
replaced the old PG *brangiana- still found in West Germanic (OE brengan w.v. I ‘bring’
OFris. brendza w.v. I ‘id.’ OS brengian w.v. I ‘id.’ OHG brengen w.v. I ‘id.’). Only later
was a full paradigm of strong forms created in analogy with other verbs with a present
stem of the structure CeRC-, i.e. other class III strong verbs. Given the validity of this
assumption, the i-stem PG *brungi- must be regarded as secondary. Literature:
Bammesberger (1990: 131), Bjorvand & Lindeman (2000: 102-103), Boutkan & Siebinga
(2005: 58-59), Holthausen (1974: 34-35), IEW (2005: 168), Kluge/Seebold (2002: 151),
Kroonen (2013: 77), Lehmann (1986: 79-80), LIV (2001: 282-284), Lloyd et al. (1998:
338-341), Orel (2003: 55-56), Philippa et al. (2003: 377), Seebold (1970: 136-137), Sehrt
(1966: 62-63).
36 PG *bruni- ‘fire flame’. Represented in OE bryne m. ‘id.’. Derivationally matching the
class III strong verb PG *brinnana- s.v. ‘burn’ > Goth. brinnan s.v. ‘id.’ ON brinna,
brenna s.v. ‘id.’ (the latter form < PG *brannjana-), OE beornan, biornan, biernan s.v.
‘id.’ OFris. berna, burna, barna w.v. I/s.v. ‘id.’ OS brinnan s.v. ’id.’ OHG brinnan s.v.
‘id.’. Further etymology uncertain but several scholars assume that a nasal present PIE
*bhre-n- H - of the root PIE *bhre H- ‘burn be wild’ also found in e.g. Lat. ferve v.
‘boil seethe’ and PG *brewwana- s.v. ‘brew’ is no inept candidate for a precursor of PG
*brinnana-. Seebold (1980: 478-479) has presented an alternative view, though, viz. that
PG *brinnana- has been created analogically from the zero grade form PG *brunn- <
*burnn- < PIE *gwhr-n - to the root PIE *gwher- ‘become warm’ cf. the cognate -neu-/-nu-
present of Skt. g ot v. lights burns’ and further e.g. Gr. θέρομαι v. ‘become warm,
warm oneself’ Lat. formus adj. ‘warm’ fornus m. ‘oven’ OIr. (fo-)geir v. ‘warms up
heats’ Lith. gar t v. ‘burn’ OCS gor t v. ’id.’. Regardless of which etymological
proposal, if any, is correct, no doubt can remain that PG *bruni- has been created to a
secondary stem rather than directly to the root, thus being of no value as a primary i-stem.

70
Literature: Bammesberger (1990: 131), Bjorvand & Lindeman (2000: 100-102), Boutkan
& Siebinga (2005: 32-33), Holthausen (1974: 24, 37), IEW (2005: 143-145),
Kluge/Seebold (2002: 149), Kroonen (2013: 77-78), Lehmann (1986: 80), LIV (2001: 96),
Lloyd et al. (1998: 342-345), Orel (2003: 56), Seebold (1970: 137-138), Sehrt (1966: 63),
De Vries (1962: 56-57).
37 PG *brusti- m. ‘break fracture’. Represented in OE byrst m. ‘loss damage’ OS (erth-)
burst m. ‘breakthrough through the earth’ OHG brust f. ‘crack fissure’. Derivationally
matching the class III strong verb PG *brestana- s.v. ‘burst break crack’ > ON bresta s.v.
‘id.’ OE berstan s.v. ‘id.’ OFris. bersta s.v. ‘break disappear’ OS brestan s.v. ‘burst
break crack’ OHG brestan s.v. ‘burst tear lack’. Further etymology uncertain except for
the obvious relationship of PG *brusti- with OIr. brissim v. ‘break’ and maybe MIr. brosc
m. ‘noise’. Literature: Bammesberger (1990: 131), Holthausen (1974: 21, 41), IEW (2005:
169), Kluge/Seebold (2002: 112), Kroonen (2013: 75), Lloyd et al. (1998: 319-322, 402),
Orel (2003: 56), Philippa et al. (2003: 228), Seebold (1970: 139), Sehrt (1966: 62-63), De
Vries (1962: 56).
38 PG *drunki- m. ‘drink’. Represented in ON drykkr m. ‘id.’ OE drync m. ‘drink swallow
gulp’ OHG trunk m. ‘drink; swallow gulp’. Derivationally matching the class III strong
verb PG *drinkana- s.v. ‘drink’ > Goth. drigkan s.v. ‘id.’ ON drekka s.v. ‘id.’ OE
drincan s.v. ‘drink swallow gulp’ OFris. drinka s.v. ‘drink’ OS drinkan s.v. ‘id.’ OHG
trinkan s.v. ‘id.’. A competing i-stem noun with radical o-grade also exists, viz. PG
*dranki- m. ‘drink’. Further etymology uncertain. Many scholars tentatively assume PG
*drinkana- to descend from the root PIE *dhreĝ- ‘glide move’ cf. e.g. Skt. dhrájati v.
‘glides slips; migrates (of birds etc.)’ and Lith. drež t v. ‘smooth straighten’ by means of
an infigated nasal and with the semantic development known from G Zug m. ‘pull draw’ >
‘swallow gulp’. Appurtenance of PG *drinkana- to Lith. dr gt v. ‘become moist become
damp’ dr gnas adj. ‘humid wet’ also by means of an infigated nasal is an option as well.
Regardless of which etymological proposal is correct, there is no doubt that PG *drunki-
has been created secondarily from a nasal present rather than directly from the root, thus
being of no value as a primary i-stem. Literature: Bammesberger (1990: 131), Bjorvand &
Lindeman (2000: 151), Holthausen (1974: 77, 79), IEW (2005: 273), Kluge/Seebold
(2002: 930), Kroonen (2013: 103), Lehmann (1986: 94-95), Orel (2003: 75, 77), Philippa

71
et al. (2003: 631, 634), Seebold (1970: 165-166), Sehrt (1966: 83-84), De Vries (1962: 82,
85).
39 PG *dunti- m. ‘shot’. Represented in ON dyttr m. ‘shot’ OE dynt m. ‘blow stroke’.
Derivationally matching the class III strong verb PG *dintana- s.v. ‘fall heavily drop’ >
ON detta s.v. ‘id.’. Further etymology uncertain. Only possible point of extra-Germanic
reference is Alb. g-dhent v. ‘cut wood plane; thrash’. Literature: Bammesberger (1990:
131), Holthausen (1974: 82), IEW (2005: 249-250), Orel (2003: 71, 79), Seebold (1970:
154), De Vries (1962: 76, 91).
40 PG *hwurbi- m. ‘way passage’. Represented in OE hwyrf m. ‘id.’. Derivationally
matching the class III strong verb PG *hwerbana- s.v. ‘walk turn’ > Goth. ƕa rban s.v.
‘go around walk; walk through life live’ ON hverfa s.v. ‘turn; disappear’ OE hweorfan
s.v. ‘turn change go’ OFris. hwerva s.v. ‘turn walk swap’ OS hwerban s.v. ‘walk hither
and thither turn return’ OHG werban, hwerban s.v. ‘return drive’. To be reconstructed as
PIE *kwr -í-to the root PIE *kwerp- ‘turn’ also found in e.g. Gr. καρπός ‘wrist i.e. where
the hand turns’ (< PIE *kwr -ó-). It should be noted that the strong verb PG *hwerbana-
and all other Germanic derivatives of that root almost consistently and deviatingly select
the voiced Verner’s variant of the final radical consonant PIE *p, i.e. PG *hwerbana- <
PIE *kwerp-é-. In other words, no forms but some Old High German variants demand the
unvoiced Verner’s variant. Literature: Bammesberger (1990: 131), Bjorvand & Lindeman
(2000: 1043-1044), Holthausen (1974: 181), IEW (2005: 631), Kluge/Seebold (2002: 983-
984), Kroonen (2013: 265-266), Lehmann (1986: 197), LIV (2001: 392-393), Orel (2003:
200), Philippa et al. (2009: 615), Seebold (1970: 282-284), Sehrt (1966: 286-287), De
Vries (1962: 271).
41 PG *kurbi- m. ‘fragment bit piece’. Represented in ON kurfr m. ‘wooden piece’ OE cyrf
m. ‘fragment bit; cutting off’ OFris. kerf m. ‘cut’. Derivationally matching the class III
strong verb PG *kerbana- s.v. ‘cut into’ > OE ceorfan s.v. ‘cut off carve’ OFris. kerva
s.v. ‘notch’. From both a formal and a semantic point of view the closest possible extra-
Germanic comparanda are constituted by Gr. γράφω v. ‘carve write’ (< PIE *grbh-) and
further OCS žr bъ m. ‘lot’ (< PIE *gerbh-o-) with cognates. Given these comparanda, PG
*kurbi- is to be reconstructed as PIE *grbh-í- to the root PIE *gerbh- ‘carve cut into’.
Literature: Bammesberger (1990: 131), Boutkan & Siebinga (2005: 214), Holthausen
(1974: 46, 68), IEW (2005: 392), Kluge/Seebold (2002: 484), Kroonen (2013: 285), LIV

72
(2001: 187), Orel (2003: 213, 225), Philippa et al. (2007: 49), Seebold (1970: 292-293), De
Vries (1962: 335).
42 PG *ru(n)ni- m. ‘run course’. Represented in Goth. runs m. (when compounded: f.) ‘race
course flow (of blood)’ ON runnr m. ‘bush grove’ OE ryne m. ‘course river run’
OFris. -rene m. ‘act of flowing’ OHG run m. ‘id.’. Radical zero grade also in e.g. PG
*runa- n. ‘flow water’ > ON run n. ‘water between two seas’ OHG run n. ‘id.’.
Derivationally matching the class III strong verb PG *rinnana- s.v. ‘run’ > Goth. rinnan
s.v. ‘run rush’ ON rinna, renna s.v. ‘run flow’ OE iernan, iornan, rinnan s.v. ‘run’
OFris. rinna s.v. ‘run go’ OS rinnan s.v. ‘flow run’ OHG rinnan s.v. ‘flow’. The strong
verb PG *rinnana- is normally compared to the -neu-/-nu-present stem of Skt. ar van v.
‘ran’ (< PIE *h3r -n -é-) which, however, according to LIV (2001: 306 with lit.) is a
scribal error for ar an. The proper comparandum of PG *rinnana- is thus not Skt. ar van
but rather Skt. r t v. ‘makes whirl sets in movement’ Gr. ὀρίνω v. ‘stir whirl; make
angry’ (< PIE *h3ri-né-H- ~ *h3ri-n-H-), i.e. a nasal present to the root PIE *h3re H-
‘bubble whirl’. Consequently PG *rinnana- < PIE *h3ri-n-H- constitutes one of the
clearest instances of gemination of resonants when originally followed by a laryngeal as
suggested by Lühr (1976: 78). Seeing that the Germanic verb must have been secondarily
adapted to the strong verbs the “zero grade” of PG *run(n)- cannot be of Indo-European
pedigree; neither can, therefore, the i-stem PG *run(n)i-. Literature: Bammesberger (1990:
131), Bjorvand & Lindeman (2000: 720-721), Boutkan & Siebinga (2005: 317-318),
Casaretto (2004: 176), Holthausen (1974: 261, 265), IEW (2005: 326-332, esp. 330-331),
Kluge/Seebold (2002: 766), Kroonen (2013: 413-414, 418), Lehmann (1986: 285-286,
288), LIV (2001: 305-306), Orel (2003: 302), Seebold (1970: 375-377), Sehrt (1966: 439),
De Vries (1962: 441-442, 453-454).
43 PG *sprungi- m. ‘jump’. Represented in OE -)spryng m. ‘id.’ OHG sprung m. ‘jump’.
Radical zero grade also in e.g. PG *s r ng n- f. ‘crack scratch’ > ON sprunga f. ‘crack in
the skin scratch’. Derivationally matching the class III strong verb PG *springana- s.v.
‘break out leap’ > ON springa s.v. ‘spring; burst crack’ OE springan s.v. ‘spring burst
out’ OFris. springa s.v. ‘spring jump leap run’ OS springan s.v. ‘id.’ OHG springan
s.v. ‘jump gush flow’. Further etymology uncertain. The closest possible cognates are Av.
- s әrәzat v. ‘strove after’ (< PIE * s rgh-), Gr. σπέρχομαι v. ‘set in rapid movement’
(< PIE *(s)pergh-) and maybe Lith. dial. re gt v. ‘press push squeeze’ (< PIE *(s)pre-n-

73
gh-) in which case the Germanic strong verb must be analysed as continuing a nasal present
PIE *s r-n-gh- with probably analogical full grade PIE *spre-n-gh- > PG *springana-. The
i-stem PG *sprungi- must consequently be seen as secondary since it has been created to
the nasal present stem rather than to the bare root. Literature: Bammesberger (1990: 131),
Boutkan & Siebinga (2005: 364-365), Holthausen (1974: 313-314), IEW (2005: 998),
Kluge/Seebold (2002: 871), Kroonen (2013: 470), LIV (2001: 581), Orel (2003: 366-367),
Philippa et al. (2009: 250), Seebold (1970: 457, 458), Sehrt (1966: 500), De Vries (1962:
538-539).
44 PG *stungi- m. ‘prick stab’. Represented in OE styng m. ‘id.’ OHG stung* m. ‘id.’ (only
attested in the passage in slegio des stunges ‘in icto puncti’). Radical zero grade also in e.g.
PG *st ng n- f. ‘stab wound’ > ON stunga f. ‘id.’. Derivationally matching the class III
strong verb PG *stingana- s.v. ‘sting stick’ > Goth. (us-)stagg s.v. ‘pluck out’ (ipv. if to
be amended to us-stigg), ON stinga s.v. ‘id.’ OE stingan s.v. ‘stick penetrate’. A
competing i-stem noun with radical o-grade also exists, viz. PG *stangi- m. ‘bar pole
stick’. Further etymology uncertain, but the Germanic strong verb could be analysed as a
nasal present PIE *st-n-gh- with analogical full grade PIE *ste-n-gh- > PG *stingana- to the
root PIE *stegh- ‘stick’ cf. also e.g. OCS o- stegnot v. ‘bind put in chains’. The i-stem
PG *stungi- should consequently be seen as secondary since it has been created to the
nasal present stem rather than to the bare root. Literature: Bammesberger (1990: 131),
Holthausen (1974: 323, 328), IEW (2005: 1014-1015), Kroonen (2013: 480), Lehmann
(1986: 383), LIV (2001: 589), Orel (2003: 374-375), Seebold (1970: 470-471), De Vries
(1962: 548, 556).
45 PG *stunkwi- m. ‘smell’. Represented in OS stunc m. ‘smell’. Radical zero grade also in
e.g. PG *stunkwa- n. ‘thrust’ > Goth. (bi-)stugq n. ‘id.’. Derivationally matching the class
III strong verb PG *stinkwana- s.v. ‘thrust clash; stink’ > Goth. stigqan s.v. ‘hit battle
stumble’ ON støkkva s.v. ‘leap jump flee’ OE stincan s.v. ‘smell stink; leap spring’ OS
stinkan s.v. ‘id.’ OHG stinkan s.v. ‘id.’. A competing i-stem noun with radical o-grade
also exists, viz. PG *stankwi- m. ‘smell’. Partly due to their very broad semantics, the
further etymology of PG *stunkwi- and *stinkwana- is highly disputed. Literature:
Bammesberger (1990: 131), Casaretto (2004: 79), Holthausen (1974: 323), IEW (2005:
1033), Kluge/Seebold (2002: 885), Kroonen (2013: 480-481, 487), Lehmann (1986: 73,

74
325), LIV (2001: 596-597), Orel (2003: 375), Philippa et al. (2009: 285), Seebold (1970:
471-472), De Vries (1962: 558).
46 PG *sturki- m. ‘strength’. Represented in ON styrkr m. ‘id.’. Radical zero grade also in e.g.
PG *sturka- m. ‘stork’ (< ‘stiff bird’) > ON storkr m. ‘id.’ OE storc m. ‘id.’ OS stork m.
‘id.’ OHG storh, storah m. ‘id.’. Derivationally matching the class III strong verb PG
*sterkana- s.v. ‘stiffen harden’ preserved in ON (blóð-)storkinn adj./ptc. ‘coagulated’ and
also confirmed by the existence of the related inchoative PG *sturknana- w.v. IV ‘become
stiff stiffen (intr.)’. To be reconstructed as PIE *strg-í- to the root PIE *sterg- ‘stiff’ cf.
also e.g. Lith. str gt v. ‘stiffen turn into ice’. Literature: Bjorvand & Lindeman (2000:
864-865), Holthausen (1974: 324), IEW (2005: 1022-1027, esp. 1023), Kluge/Seebold
(2002: 887), Kroonen (2013: 488), Orel (2003: 375, 384), Seebold (1970: 473-474), De
Vries (1962: 551, 558).
47 PG *swulgi- m. ‘drink swallow gulp’. Represented in ON sylgr m. ‘drink, swallow, gulp;
drinker swallower’ i.e. both as abstract noun and agent noun. Radical zero grade also in
e.g. PG *swulga- m. ‘whirl whirlpool vortex’ > OS swolg m. ‘id.’. Derivationally
matching the class III strong verb PG *swelgana- s.v. ‘swallow’ > ON svelga s.v.
‘swallow choke on’ OE swelgan s.v. ‘swallow’ OHG swelgan, swel(a)han s.v. ‘id.’. A
competing i-stem noun with radical o-grade also exists, viz. PG *swalgi- m. ‘abyss; swirl
whirlpool’. Further etymology uncertain. In my view the best proposal is the one by
Specht (1939: 25-26), viz. that PG *swelgana- (< PIE *s elk-) is to be analysed as a
contamination into PIE *s elk- of PIE *h2 elk- ‘pull drag’ on the one hand cf. e.g. Gr.
αὖλαξ f. ‘furrow’ (< PIE *h2 lk- < *h2ulk- with syllabification based analogically on the o-
grade represented in Gr.(Hom.) acc.sg. ὦλκα < *ἄϝολκα < PIE *h2 lk- , cf. Schindler
(1972: 34)), Lith. v lkt v. ‘id.’ and PIE *selk- ‘pull’ on the other cf. Arm. ełg adj. ‘slow
inert’ Gr. ἕλκω v. ‘pull’ Gr. ὁλκή f. ‘drawing dragging tugging’ ὁλκός m. ‘hauling-
machine; furrow’ Alb. helq v. ‘pull tear off’ Lat. sulcus m. ‘furrow’ s lc v. ‘plough’
and further PG *sulh- f. ‘plough; furrow’ and *selha- m. ‘seal (animal) that drags itself
along the ground’. As with PG *drunki- m. ‘drink’, the semantic development would be
the one known from G Zug m. ‘pull draw’ > ‘swallow gulp’. Literature: Bammesberger
(1990: 132), Bjorvand & Lindeman (2000: 882), Holthausen (1974: 334), IEW (2005: 901,
1145), Kluge/Seebold (2002: 833), LIV (2001: 289-290, 530-531, though no mentioning of

75
the Germanic forms), Orel (2003: 394), Seebold (1970: 488-489), De Vries (1962: 567,
573).
48 PG *swuli- m. ‘swelling’. Represented in OE swile, swyle m. ‘tumor; callosity callous
skin’. Radical zero grade also in e.g. PG *swulla- m. ‘swelling’ > ON sullr m. ‘tumor’
OFris. swoll m. ‘id.’. Derivationally matching the class III strong verb PG *swellana- s.v.
‘swell’ > ON svella s.v. ‘id.’ OE swellan s.v. ‘id.’ OFris. swella, swilla s.v. ‘swell; rise
stand up’ OS swellan s.v. ‘swell’ OHG swellan s.v. ‘id.’. A competing i-stem noun with
radical full grade also exists, viz. PG *sweli- m. ‘callosity callous skin’. To be
reconstructed as PIE *s lH-í- to the root PIE *s elH- ‘swell’ whose final laryngeal is
suggested by Lühr (1976: 92) on the basis of the geminated *l. Only one possible extra-
Germanic comparandum exists, viz. Lat. īn- sol sc v. ‘become conceited become rude;
swell’ (< PIE *s lH-eh1-). Finally it should be noted that since the attestation of PG
*swuli- m. ‘tumor; callosity callous skin’ is limited to OE swyle m. ‘id.’ which if a free
variant of swile, may alternatively be reconstructed as PG *sweli-, there is no real basis for
the reconstruction of PG *swuli-. Literature: Bammesberger (1990: 133), Boutkan &
Siebinga (2005: 386-387), Holthausen (1974: 334-335, 337), Kluge/Seebold (2002: 833),
Kroonen (2013: 499, 503), LIV (2001: 609-610), Orel (2003: 394, 398), Philippa et al.
(2009: 683), Seebold (1970: 489-490), De Vries (1962: 560, 567).
49 PG *swulti- m. ‘starvation’. Represented in OE swylt m. ‘death’. Radical zero grade also in
e.g. PG *swulta- m. ‘starvation’ > Goth. swulta(-wairþja) m. ‘person near death’ ON sultr
m. ‘hunger’. Derivationally matching the class III strong verb PG *sweltana- s.v. ‘dwindle
starve perish’ > Goth. swiltan s.v. ‘die lie dying’ ON svelta ‘starve suffer hunger; die’
OE sweltan, swiltan, swyltan s.v. ‘perish’ OS sweltan s.v. ‘die’ OHG swelzan s.v. ‘burn’.
To be reconstructed as PIE *s ld-í- to the possibly extended root PIE *s el-d- ‘singe burn;
smart’ cf. also e.g. Lith. ( emaitic) sv lt v. ‘smoulder singe; smart’ and PG *swelana-
s.v. ‘singe; smart’ (> OE swelan s.v. ‘suppurate’). Literature: Bammesberger (1990: 132),
Bjorvand & Lindeman (2000: 875-876, 883), Holthausen (1974: 335, 339), IEW (2005:
1045), Kroonen (2013: 499, 503), Lehmann (1986: 336), LIV (2001: 610), Orel (2003:
394, 398), Seebold (1970: 490-491), Sehrt (1966: 517), De Vries (1962: 560, 567-568).
50 PG *wunni- m. ‘suffering pain’. Represented in Goth. wunns m.(?) ‘id.’. Derivationally
matching the class III strong verb PG *winnana- s.v. ‘exert oneself take pains’ > Goth.
winnan s.v. ‘suffer’ ON vinna s.v. ‘work do complete; fight win’ OE winnan s.v. ‘fight,

76
suffer reach’ OFris. winna s.v. ‘gain win get’ OS winnan ‘win gain; rage’ OHG
winnan s.v. ‘exert oneself take pains; work; fight rage’. Competing i-stem nouns with
radical full and lengthened grade also exist, viz. PG *weni- m. ‘friend’ and PG *w n - m.
‘hope expectation’. There is little doubt that the precursor of PG *winnana- and *wunni-
is the root PIE * enH- ‘strive want desire’ cf. also e.g. Skt. vanóti v. ‘loves; wants
demands; wins’ vánati v. ‘id.’ v tá- adj./ptc. ‘loved; wanted demanded; won’ (< PIE
* nH-tó-), vanita- adj./ptc. ‘id.’ Lat. venus f. ‘(goddess of) love pleasure grace’ (gen.sg.
veneris), OIr. fine f. ‘family kin’ (< PIE * en- e 2- < * enH- e 2-; for the development of
PIE *- H - > *- - cf. Pinault (1982: 266)), OCS uniti v. ‘will want’ and maybe Hitt.
wenzi v. ‘sleeps with’. The geminated *n of the Germanic forms is to be explained as
arisen from either of the sequences nasal followed by laryngeal, cf. tentatively Lühr (1976:
80-81), or PIE *-n -, cf. the -neu-/-nu-present of Skt. vanóti (< PIE * n-n - ~ * n-nu-)
which, if thematisised (i.e. < PIE * n-nu-é-), would yield PG *wunnana- to which a new
present stem *winnana- could easily be shaped in analogy with other class III strong verbs.
Literature: Bammesberger (1990: 132), Bjorvand & Lindeman (2000: 1057-1058),
Boutkan & Siebinga (2005: 451), Casaretto (2004: 178), Holthausen (1974: 398), IEW
(2005: 1146-1147), Kluge/Seebold (2002: 356), Kroonen (2013: 587-588), Lehmann
(1986: 404), LIV (2001: 682-683), Orel (2003: 682-683), Philippa et al. (2009: 629),
Seebold (1970: 556-557), Sehrt (1966: 699-700), De Vries (1962: 666).
51 PG *wurdi- f. ‘fate destiny chance’. Represented in ON urðr f. ‘(goddess of) fate
destiny’ m. ‘bad luck death’ OE wyrd f. ‘fate destiny’ OS wurd f. ‘id.’ OHG wurt f.
‘id.’. Derivationally matching the class III strong verb PG *werþana- s.v. ‘become appear
come into being come about happen’ > Goth. wairþan s.v. ‘id.’ ON verða s.v. ‘id.’ OE
weorthan s.v. ‘id.’ OFris. wertha s.v. ‘become be assigned’ OS werthan, werdan s.v.
‘become appear come into being happen’ OHG werdan s.v. ‘id.’. To be reconstructed as
PIE * rt-í- to the root PIE * ert- ‘turn’ cf. also e.g. Skt. vártate v. ‘turns’ Lat. vert v.
‘turn change’ Lith. verst v. ‘turn overturn (pres. ver ), OCS vrьt t s v. ‘turn’; for the
semantic development cf. Eng. turn ‘turn’ > ‘turn become’ or Fr. devenir v. ‘become’ <
Lat. d ven v. ‘come down; go to arrive at reach’. The motivation for the general
feminine gender of PG *wurdi- as opposed to the masculine gender of virtually any other
primary i-stem in Germanic is unknown. We could, however, consider the possibility of
reanalysis of PG *wurd-í- as *wur-dí-, i.e. as a feminine ti-stem. Literature: Bammesberger

77
(1990: 132), Bjorvand & Lindeman (2000: 1070-1071), Boutkan & Siebinga (2005: 431,
444), Holthausen (1974: 390, 411), IEW (2005: 1156-1158), Kluge/Seebold (2002: 984),
Kroonen (2013: 581-582, 600), Lehmann (1986: 391), LIV (2001: 691-692), Orel (2003:
455, ), Philippa et al. (2003: 636-637), Seebold (1970: 559-561), Sehrt (1966: 653-659),
De Vries (1962: 635-636, 655).

2.1.4. I-stem verbal abstracts corresponding to class IV strong verbs

52 PG *bruki- m. ‘breach crack fragment’. Represented in OE bryce m. ‘id.’ OFris. breke,

breze m./f. ‘id.’ OS bruki- m. ‘id.’ OHG bruc m. ‘id.’. Radical zero grade also in e.g. PG
*bruka- m. ‘breaker’ > OHG (pi-)proh m. ‘seducer’ and PG *bruka- n. ‘fragment’ > OE
(ge-)broc n. ‘id.’. Derivationally matching the class IV strong verb PG *brekana- s.v.
‘break crack’ > Goth. brikan s.v. ‘fight; break destroy’ OE brecan s.v. ‘break storm’
OFris. breka s.v. ‘break tear’ OS brekan s.v. ‘id.’ OHG brechan, brehhan s.v. ‘break;
split open’. Extra-Germanic comparanda are few and uncertain, but Lat. frang v. ‘break’
is the one most often mentioned and also the most plausible candidate if we can assume
with Schrijver (1991: 478) that Lat. frang < PIE *bhre-n-ĝ-, i.e. a reduced grade as
replacement of the proper zero grade PIE *bhrĝ- > Lat. †forg- vel sim. in a way quite
similar to the one in which the original zero grade PG *burk- (< PIE *bhrĝ-) was altered
into PG *bruk- in order to maintain the phonotactics of the root PIE *bhreĝ-. If OIr.
braigid v. ‘farts’ is to be compared here as well a similar reduced grade must be
reconstructed for its precursor. Literature: Bammesberger (1990: 133), Bjorvand &
Lindeman (2000: 99-100), Boutkan & Siebinga (2003: 61-62), Holthausen (1974: 33, 35-
36), IEW (2005: 165), Kluge/Seebold (2002: 148, 153), Kroonen (2013: 75, 79), Lehmann
(1986: 80), LIV (2001: 91-92), Lloyd et al. (1998: 307-309, 374), Orel (2003: 55), Philippa
et al. (2003: 376-377), Seebold (1970: 132-135), Sehrt (1966: 61-62).
53 PG *buri- m. ‘son progeny’. Represented in Goth. baur m. ‘son’ ON burr m. ‘id.’ OE
byre ‘son progeny; time occasion’. Radical zero grade also in e.g. PG *bura- adj. ‘high’ >
OHG bor adj. ‘id.’ (only in in bor(e) ‘upwards in the air’) PG *bura- m. ‘party’ > Goth.
(ga-)baur m. ‘id.’; and PG *bura- n. ‘collect collection; progeny’ > Goth. (ga-)baur n.
‘collect collection’ OHG (gi-)por n. ‘progeny’. Derivationally matching the class IV
strong verb PG *berana- s.v. ‘carry; bear give birth’ > Goth. bairan s.v. ‘id.’ ON bera
s.v. ‘id.’ OE beran s.v. ‘carry; tolerate’ OS beran s.v. ‘carry possess’ OHG beran s.v.
‘produce bear’. To be reconstructed somewhat anachronistically as *bhr-í- to the widely

78
attested root PIE *bher- ‘carry’ cf. also e.g. Skt. bhárati, bíbharti, bibhárti v. ‘carries’ Av.
baraiti v. ‘id.’ Arm. berem v. ‘carry’ Phryg. (αβ- βερετ v. ‘carried led’ Gr. φέρω v.
‘carry’ Lat. fer v. ‘id.’ OIr. beirid v. ‘carries’ OCS bьrat v. ‘collect read’ (pres. berǫ).
Literature: Bammesberger (1990: 132), Bjorvand & Lindeman (2000: 126), Boutkan &
Siebinga (2005: 38-39), Casaretto (2004: 53-54, 79, 174), Holthausen (1974: 21, 40), IEW
(2005: 128-132), Kluge/Seebold (2002: 335), Kroonen (2013: 59, 85), Lehmann (1986: 57,
64), LIV (2001: 76-77), Lloyd & Springer (1988: 546-548), Lloyd et al. (1998: 241-243),
Orel (2003: 41-42, 64), Philippa et al. (2003: 223), Seebold (1970: 104-106), Sehrt (1966:
44), De Vries (1962: 33, 65, 68).
54 PG *drupi- m. ‘blow stroke’. Represented in OE drype m. ‘id.’. Derivationally matching
the class IV strong verb PG *drepana- s.v. ‘blow strike’ > ON drepa s.v. ‘blow strike;
kill slay’ OE drepan s.v. ‘hit strike; kill slay’ OS -drepan s.v. ‘exceed surpass’ OHG
treffan s.v. ‘blow strike hit; punish’. A competing i-stem noun with radical o-grade also
exists, viz. PG *drapi- m. ‘blow stroke’. To be reconstructed as PIE *dhrb-í-24 or
presumably rather PIE *dhrbh-í- to the root PIE *dhrebh- ‘crush grind’ also attested in
Slavic, cf. e.g. OCS drobiti v. ‘crush break’ drobь f. ‘piece fragment’ if we follow
Kroonen’s (2013: 101-102) analysis stating that PG *drepana- has been backformed from
the related iterative verb PG *dr b (ja)na- w.v. II ‘hit’ < PIE *dhrbh-n- with PIE *-P-n-
> PG *-pp- by means of Kluge’s Law. Literature: Holthausen (1974: 77, 79; translating OE
drype with ’fall’ and comparing it to PG *dreupana- s.v. ‘drip’) IEW (2005: 272-273),
Kluge/Seebold (2002: 927), Kroonen (2013: 101-102, 105), LIV (2001: 153), Orel (2003:
75), Seebold (1970: 166-167), De Vries (1962: 83).
55 PG *kuli- m. ‘cold’. Represented in ON kylr m. ‘id.’ (reshaped as a ja-stem), OE cyle m.
‘id.’. Derivationally matching the archaic preterite participle stem of the class VI strong
verb PG *kalana- s.v. ‘be cold’ (originally probably class IV *kelana-) found only in OSw.
kolin adj./ptc. ‘feeling shivery’ (< PG *kulena- ~ *kulana- pret.ptc.). Only later was a
competing i-stem PG *kali- m. ‘cold’ created to the new preterite participle PG *kalena- ~
*kalana-. To be reconstructed as PIE *glH-í- to the root PIE *gelH- ‘cold’ cf. also e.g.
Lat. gelus m., gelu n. ‘cold frost ice’ gelidus adj. ‘cold’ Lith. gél(-menis) f. ‘severe cold’.

24 With metathesis of PG *ur > *ru in order to restore the inflectional paradigm by bringing the phonotactics in line
with the full and o-grade forms PG *drep- and *drap- of the phonotactic structure CRVC.

79
Literature: Bammesberger (1990: 132), IEW (2005: 365-366), Kroonen (2013: 277, 309),
Orel (2003: 208-209, 223), Seebold (1970: 288-289), De Vries (1962: 340).
56 PG *kwumi- m. ‘coming’. Represented in Goth. qums m. ‘coming approach’ OE cyme m.
‘coming arrival’ OFris. (on-)keme m. ‘penetration’ OS kumi m. ‘advent’ OHG (fona-uf-)
kume m. ‘arrival’. Derivationally matching the class IV strong verb PG *kwemana- ~
*k(w)umana- s.v. ‘come’ > Goth. qiman s.v. ‘id.’ ON koma s.v. ‘id.’ OE cuman s.v. ‘id.’
OFris. kuma s.v. ‘id.’ OS kuman s.v. ‘id.’ OHG queman s.v. ‘id.’. A competing i-stem
noun with a radical full grade also exists, viz. PG *kwemi- m. ‘eventus i.e. outcome,
result’. To be reconstructed as *gw -í- to the root PIE *gwem- ‘come’ cf. also e.g. Skt.
gácchati v. ‘comes’ gáti- f. ‘going’ Gr. βαίνω v. ‘come’ βάσκω v. ‘id.’ βάσις f. ‘step’
Lat. ven v. ‘come’ Lith. g t v. ‘be born’ (pres. gemù). Literature: Bammesberger
(1990: 132), Bjorvand & Lindeman (2000: 472-473), Boutkan & Siebinga (2005: 210-211,
224), Casaretto (2004: 174), Holthausen (1974: 63, 67), IEW (2005: 464-465),
Kluge/Seebold (2002: 513), Kroonen (2013: 316), Lehmann (1986: 276-277, 279), LIV
(2001: 209-210), Orel (2003: 227-228, 231), Philippa et al. (2007: 108), Seebold (1970:
315-317), Sehrt (1966: 312-315), De Vries (1962: 325).
57 PG *muni- m. ‘thought feeling’. Represented in Goth. muns m. ‘thought intension’ ON
munr m. ‘mind longing passion desire’ OE myne m. ‘mind thought purpose’.
Derivationally matching the class IV preterite-present verb PG *man ~ *munum pp.v.
‘think’ > Goth. man ~ munum pp.v. ‘think believe regard’ ON man ~ munum pp.v.
‘remember mark’ OE man ~ *munun pp.v. ‘regard’ OS (far-)man ~ (far-)munum pp.v.
‘despise deny’. To be reconstructed as * n-í- to the root PIE *men- ‘think’ cf. also e.g.
Skt. mányate v. ‘thinks regards’ manas- n. ‘mind’ Av. manah- n. ‘id.’ Lith. miñti v.
‘think’ (pres. menù), OCS ьn t v. ‘id.’ (pres. ьnjǫ) and the perfects of Av. mamne v.
‘has thought’ Gr. μέμονα v. ‘have in mind’ Lat. e nī v. ‘think of remember mention’
to which PG *man ~ *munum serves as an exact cognate. Literature: Bammesberger (1990:
132-133), Bjorvand & Lindeman (2000: 613), Casaretto (2004: 175), Holthausen (1974:
227-228), IEW (2005: 726-728), Kluge/Seebold (2002: 590), Kroonen (2013: 375-376),
Lehmann (1986: 260-261), LIV (2001: 435-436), Orel (2003: 259, 276), Seebold (1970:
345-346), Sehrt (1966: 400), De Vries (1962: 396).
58 PG *numi- m. ‘taking’. Represented in OE (fore-)nyme m. ‘praesumptio i.e. anticipation’.
Radical zero grade also in e.g. PG *numan- m. ‘taker; taken(?)’ > ON (her-)numi m.

80
‘prisoner of war’ OE (ierfe-)numa m. ‘heir’ OFris. (erf-)noma m. ‘id.’ OS (sigi-)nomo m.
‘victor’. Derivationally matching the class IV strong verb PG *nemana- s.v. ‘take’ > Goth.
niman s.v. ‘take catch’ ON nema s.v. ‘take catch; learn’ OE niman s.v. ‘take’ OFris.
nema, nima s.v. ‘take rob’ OS neman s.v. ‘take choose’ OHG neman s.v. ‘take grab’.
An i-stem adjective with a radical lengthened grade also exists, viz. PG *n - adj. ‘easy
to take acceptable’. To be reconstructed as *n -í- to the root PIE *nem- ‘assign to allot
to; take; give’ also found in e.g. Av. nә a - n. ‘loan’ Gr. νέμω v. ‘distribute; seize
possess’ Lat. numerus m. ‘number’ (< PIE *nomeso-), Lith. núoma f. ‘rent tenancy’.
Latv. e t v. ‘take’. Literature: Bammesberger (1990: 133), Bjorvand & Lindeman (2000:
653-655), Boutkan & Siebinga (2005: 281, 286-287), Holthausen (1974: 236, 239), IEW
(2005: 763-764), Kluge/Seebold (2002: 648), Kroonen (2013: 387), Lehmann (1986: 266-
267), LIV (2001: 433), Orel (2003: 284), Philippa et al. (2007: 414), Seebold (1970: 357-
359), Sehrt (1966: 412-413), De Vries (1962: 407).
59 PG *stuni- m. ‘groan’. Represented in ON stynr m. ‘id.’ (reshaped as a ja-stem possibly
due to influence from ON stynja w.v. I ‘groan’ < PG *stunjana-). Derivationally matching
the class IV strong verb PG *stenana- s.v. ‘groan’ > OE stenan s.v. ‘roar groan’. To be
reconstructed somewhat anachronistically as PIE *stn-í- to the root PIE *sten- ‘roar
produce unarticulated sounds’ found also in Greek and Balto-Slavic, viz. in Gr. στένω v.
‘moan sigh’ στόνος m. ‘sigh’ Lith. sten t v. ‘moan’ (pres. stenù), OCS stenati v. ‘id.’
(pres. stenjǫ). Literature: Bammesberger (1990: 133), Bjorvand & Lindeman (2000: 870),
Holthausen (1974: 319), IEW (2005: 1021), Kluge/Seebold (2002: 886), LIV (2001: 596),
Orel (2003: 374), Seebold (1970: 469-470), De Vries (1962: 557).

2.1.5. I-stem verbal abstracts corresponding to class V strong verbs

60 PG *kwedi- m. ‘talk’. Represented in OE cwide m. ‘word saying; order command; curse’

OS quidi m. ‘word’ OHG quiti f. ‘pronunciation’. Radical full grade also in e.g. PG
*kwedu- m. ‘talk’ > ON kviðr m. ‘saying word; verdict inquest’. Derivationally matching
the class V strong verb PG *kweþana- s.v. ‘talk’ > Goth. qiþan s.v. ‘say speak’ ON kveða
s.v. ‘say speak; sing recite’ OE cwethan s.v. ‘say speak’ OFris. quetha s.v. ‘say speak;
mean’ OS quethan s.v. ‘say speak’ OHG quedan s.v. ‘say; ensure’. To be reconstructed
as PIE *gwet-í- to the root PIE *gwet- ‘say speak’ the prevalence of which in other
branches is disputed. Skt. gádati v. ‘speaks’ (with *t replaced by d in analogy with Skt.
vádati v. ‘speaks’) Arm. ko cem v. ‘call invite’ and Lat. vet v. ‘forbid’ (< ‘say no’) are

81
often mentioned as possible cognates of PG *kweþana-, but for the Latin form to be thus
compared, it must first be separated from OW guetid v. ‘says’ (< PIE * et 2-). Literature:
Bammesberger (1990: 136), Bjorvand & Lindeman (2000: 488), Boutkan & Siebinga
(2005: 312), Holthausen (1974: 65), IEW (2005: 480-481), Kroonen (2013: 315, 319),
Lehmann (1986: 277-278), LIV (2001: 212), Orel (2003: 229), Seebold (1970: 318-320),
Sehrt (1966: 429-432), De Vries (1962: 338).
61 PG *wreki- m. ‘pursuit’. Represented in OHG (ge-)rih m. ‘avenge punishment’. Radical
full grade also in e.g. PG *wreka- n. ‘sth. drifting’ > ON rek n. ‘driftwood’. Derivationally
matching the class V(/IV) strong verb PG *wrekana- s.v. ‘chase pursue; drive out oust’ >
Goth. wrikan s.v. ‘chase pursue’ ON reka s.v. ‘drive carry out; hunt revenge’ OE
wrecan s.v. ‘drive out oust; avenge’ OFris. wreka s.v. ‘pursue; avenge’ OS wrekan s.v.
‘pay back avenge’ OHG rehhan s.v. ‘pursue; punish avenge’. To be reconstructed as PIE
* reg-í- to the root PIE * reg- ‘walk drive; drive off push; follow a lead’ cf. also e.g.
Hitt. ūrk - c. ‘trace’ Skt. vrájati v. ‘walks strides’ ar -v j- m. ‘outlaw wreck
miserable’ Lat. rge v. ‘press oppress’ OCS vr t v. ‘throw’ (pres. vrьgǫ). Literature:
Bammesberger (1990: 137), Bjorvand & Lindeman (2000: 719), Boutkan & Siebinga
(2005: 459), Holthausen (1974: 407), IEW (2005: 1181), Kluge/Seebold (2002: 738),
Kroonen (2013: 595), Lehmann (1986: 410), LIV (2001: 697), Orel (2003: 471-472),
Philippa et al. (2009: 641), Seebold (1970: 568-570), Sehrt (1966: 721), De Vries (1962:
440).

2.1.6. I-stem verbal abstracts corresponding to class VI strong verbs

62 PG *agi- m. ‘fear’. Represented in OE ege m. ‘id.’. Identical radical ablaut grade in e.g.
PG *agan- m. ‘fear horror; unrest’ > ON agi m. ‘id.’; PG *agīn- f. ‘fear’ > Goth. (un-
)agein adv. ‘fearless’ OHG egī f. ‘id.’; and PG *agiz- n. ‘fear’ > Goth. agis n. ‘id.’ (a-
stem) and, with a secondary n-stem enlargment in West Germanic, OE egesa m. ‘horror’
OS egiso m. ‘id.’ OHG agiso, egiso m. ‘id.’. Derivationally matching the class VI
preterite-present verb PG * g ~ * g pp.v. ‘fear’ > Goth. og ~ ogum pp.v. ‘id.’. The
original preterite participle of this verb is not attested, the verb being a preterite-present;
the secondary weak preterite participle is accidentally unattested as well. The original a-
vocalism normally prevailing in the present stem and the preterite participle stem of class
VI strong verbs is secured, though, by the fossilised present participle Goth. (un-)agands
adj. ‘fearless’ i.e. ‘not fearing’ with the varia lectio of un-agans which is, in the light of

82
the semantics (‘fearless’ = ‘not fearing’ rather than ‘not feared’) probably to be seen as a
mere scribal error for un-agands rather than as a true preterite participle. Outside
Germanic, an exact cognate to the s-stem prevails in Gr. ἄχος n. ‘sadness pain’. Further
comparanda include Skt. aghá- n. ‘pain danger’ Gr. ἄχνυμαι, ἄχομαι v. ‘grieve mourn
am sad’ and OIr. (ad-)ágathar v. ‘fears’ and the root is thus to be reconstructed as PIE
*h2egh- ‘fear’. Besides being analysable as an i-stem verbal abstract formed automatically
from the ablaut grade found in the weak stem of the original preterite participle of the
corresponding verb, PG *agi- may also be regarded as continuing the s-stem PIE *h2égh-os
~ * h2égh-es- > PG *ag-az ~ * ag-iz-, the weak stem of which is, by means of
reinterpretation of the stem-final PG *z as the case marker of the nom.sg., easily
resegmentable as PG *ag-i-. Literature: Bammesberger (1990: 134), Casaretto (2004: 559),
Holthausen (1974: 89), IEW (2005: 7-8), Kroonen (2013: 3-4), Lehmann (1986: 9-10,
270), LIV (2001: 257), Lloyd et al. (1998: 957, 962-964), Orel (2003: 3, 290), Seebold
(1970: 362), De Vries (1962: 3).
63 PG *aki- m. ‘ache pain’. Represented in OE ece m. ‘id.’. Derivationally matching the class
VI strong verb PG *akana- s.v. ‘ache’ > OE acan s.v. ‘ache’. Since it is restricted to
English and has no certain etymology, Seebold (1970: 75) is probably right in assuming
that OE acan s.v. ‘ache’ is etymologically identical to PG *akana- s.v. ‘lead drive’ > ON
aka s.v. ‘drive’ with a long range of extra-Germanic cognates, cf. e.g. Toch. B ś v.
‘travel lead’ Skt. ījate v. ‘leads drives’ (< PIE *h2i-h2ĝ-e-), ájati v. ‘id.’ Av. azaiti v.
‘id.’ Arm. acem v. ‘lead bring’ Gr. ἄγω v. ‘lead drive’ Lat. ag v. ‘id.’ all of which
descend from the root PIE *h2eĝ- ‘lead’. A semantic parallel of PG *akana- s.v. ‘lead
drive’ vs. *akana- s.v. ‘ache pain’ is found in the intensive verb Lat. ag t v. ‘put in
motion drive’ > ‘rouse up disturb worry torment’. Literature: Bammesberger (1990:
134), Bjorvand & Lindeman (2000: 23), Holthausen (1974: 2, 87), IEW (2005: 4-6, esp. 4-
5), Kroonen (2013: 18), LIV (2001: 255-256), Orel (2003: 11), Seebold (1970: 74-75), De
Vries (1962: 3).
64 PG *kali- m. ‘cold’. Represented in OE ciel m. ‘id.’. Derivationally matching the newly
shaped preterite participle stem of the class VI strong verb PG *kalana- s.v. ‘be cold’
(originally probably class IV *kelana- in the light of the archaic-looking preterite participle
OSw. kolin adj./ptc. ‘feeling shivery’) > ON kala s.v. ‘become cold be cold be stiff’ OE
calan s.v. ‘be cold’. With PG *kali- being derived from PG *kalena- ~ *kalana- ptc.

83
‘cold’ i.e. the newly shaped preterite participle stem of PG kalana- attested in ON kalinn
ptc. ‘id.’ and OE calen ptc. ‘id.’ it must be regarded as secondary as well. Literature:
Bammesberger (1990: 134), Holthausen (1974: 42, 47), IEW (2005: 365-366), Kroonen
(2013: 277), Orel (2003: 208-209), Seebold (1970: 288-289), De Vries (1962: 297-298).
65 PG *skapi- m. ‘nature character kind’. Represented as simplex only in OHG scaf m. ‘id.’;
as a second member of compounds it is widely attested in the Germanic languages, cf. e.g.
ON -skapr m. ‘of that kind’ OE -sciepe m. ‘id.’ OFris. -skipi m. ‘id.’ OS -skepi m. ‘id.’
OHG -scaf m. ‘id.’. Radical zero grade also in e.g. PG *skapa- n. ‘vessel; nature character
kind; form creation’ > ON skap n. ‘nature character kind’ OE (ge-)sceap n. ‘creation,
form; destiny’ OS skap n. ‘vessel container’ OHG skaf n. ‘nature character kind; vessel
container’. Derivationally matching the class VI strong verb PG *skapjana- s.v. ‘form
create’ (with the present stem formed as a class I weak verb in some of the Germanic
languages) > Goth. (ga-)skapjan s.v. ‘id.’ ON skepja s.v./w.v. ‘shape form mould’ OE
scieppan, scippan s.v. ‘form create’ OFris. skeppa s.v. ‘create; destine’ OS skeppian s.v.
‘id.’ OHG scapfen, scepfen s.v./w.v. ‘id.’. To be reconstructed as PIE *skHb-í- to a root
PIE *skeHb- ‘form create’ vel sim. with no extra-Germanic cognates at first sight.
However, there might not be any need for the reconstruction of a root-final PIE *b for the
precursor of PG *skapi- and *skapjana-. Following Kroonen (2013: 440), we might
suggest that PG *p is unoriginal, having rather spread from the possibly related iterative
verb PG *ska b ja na- w.v. II ‘hollow out’ > ‘shave scrabe’ where Kluge’s Law (PIE
*-P-n- > PG *-pp-) was in operation. Consequently, any labial plosive will do, and our
minds may instantly turn towards the root PIE *skabh- or *skh2ebh- ‘scratch scrabe’ also
found in e.g. Gr. σκάφη f. ‘basin trough; hollow’ σκάπτω v. ‘dig dig up’ (with analogical
π for φ), Lat. scab v. ‘scratch rub’ Lith. skàbti v. ‘pick pluck’ and also PG *skabana-
s.v. ‘shave scrabe’. Literature: Bammesberger (1990: 135), Boutkan & Siebinga (2005:
347-348, 353), Holthausen (1974: 273, 277), IEW (2005: 930-933), Kluge/Seebold (2002:
791), Kroonen (2013: 440), Lehmann (1986: 148-149), LIV (2001: 549), Orel (2003: 334),
Philippa et al. (2009: 84), Seebold (1970: 406-408), Sehrt (1966: 469), De Vries (1962:
483, 489).
66 PG *slagi- m. ‘hit blow stroke’. Represented in Goth. slahs m. ‘slap box on the ear’ ON
slagr m. ‘hit blow stroke’ OE slege m. ‘id.’ OFris. slei m. ‘id.’ OS slegi m. ‘id.’ OHG
slag m. ‘id.’. Identical radical ablaut grade in e.g. PG *slaga- n. ‘hit blow stroke’ > ON

84
slag n. ‘id.’; PG *slag - f. ‘blow’ > OE -slagu f. ‘slag’ OS f-)slaga f. ‘track of a hoof’
OHG slaga f. ‘blow’ (only in n slago dero br wo ‘in the blink of an eye’); and PG
*slagan- m. ‘slayer’ > OE (man-)slaga m. ‘murderer man-slayer’ OS (man-)slago m.
‘id.’ OHG (man-)slago m. ‘id.’. Derivationally matching the class VI strong verb PG
*slahana- s.v. ‘hit blow beat strike slay’ > Goth. slahan s.v. ‘id.’ ON slá s.v. ‘id.’ OE
sl an s.v. ‘id.’ OFris. sl , sl n s.v. ‘id.’ OS slahan s.v. ‘id.’ OHG slahan s.v. ‘strike
smite’. To be reconstructed as PIE *slak-í- to a root PIE *slak- ‘strike’ for which Celtic
offers the only extra-Germanic comparanda, cf. e.g. MIr. slacc sb. ‘sword’ slachta
adj./ptc. ‘struck’. Literature: Bammesberger (1990: 135), Bjorvand & Lindeman (2000:
818-819), Boutkan & Siebinga (2005: 356-357), Casaretto (2004: 183), Holthausen (1974:
297-298), IEW (2005: 959), Kluge/Seebold (2002: 805-806), Kroonen (2013: 452),
Lehmann (1986: 314), LIV (2001: 564), Orel (2003: 348), Philippa et al. (2009: 165-167),
Seebold (1970: 425-427), Sehrt (1966: 478-479), De Vries (1962: 512).
67 PG *stadi- m./f. ‘place town’. Represented in Goth. staþs m. ‘place region’ ON staðr m.
‘standing; place’ OE stede m. ‘place condition’ OFris. sted m., stede f. ‘city town’ OS
stedi m. ‘place’ OHG stat f. ‘id.’. Partially contra Bammesberger (1990: 135), PG *stadi-
is probably to be regarded as a ti-stem abstract to the root PIE *steh2- ‘stand’ i.e. PIE
*st 2-tí-, cf. also e.g. Skt. sthití- f. ‘standing position’ and Gr. στάσις f. ‘id.’ rather than as
an i-stem abstract to the secondary Germanic root PG *staþ- ~ *stad- ‘stand’ reflected in
the class VI strong verb PG *standana- s.v. ‘stand’. The otherwise inexplicable feminine
gender of some of the West Germanic forms seems to support that analysis. Consequently,
PG *stadi- is of no relevance to the purpose of this study. Literature: Bammesberger
(1990: 135), Bjorvand & Lindeman (2000: 850-851), Boutkan & Siebinga (2005: 370),
Casaretto (2004: 512), Holthausen (1974: 318), IEW (2005: 1004-1010, esp. 1006),
Kluge/Seebold (2002: 873), Kroonen (2013: 472), Lehmann (1986: 323-324), LIV (2001:
590-592), Orel (2003: 369), Philippa et al. (2009: 257-258), Seebold (1970: 460-461),
Sehrt (1966: 505), De Vries (1962: 540).
68 PG *stapi- m. ‘step’. Represented in OE stæpe, stepe m. ‘id.’ OFris. -stepi m. ‘step
going’ OS stap m. ‘step’ OHG stapf m. ‘id.’. Derivationally matching the class VI strong
verb PG *stapjana- s.v. ‘walk step’ (with the present stem formed as a class I weak verb
in some of the Germanic languages) > OE stæppan s.v. ‘trudge plod’ OFris. stapa s.v.
‘walk step’ OS stapen s.v. ‘walked stepped’ (pret.ptc.). According to Kroonen (2013:

85
474), the strong verb is likely to be a backformation from the iterative PG *sta b ja na-
w.v. II ‘walk’ and thus to be reconstructed as PIE *stop-n- (> PG *stapp- by means of
Kluge’s Law) to a root PIE *step- ‘track step’ also found in e.g. OCS stopa f. ‘footprint,
sole’. Literature: Bammesberger (1990: 135), Boutkan & Siebinga (2005: 367-368, 373),
Holthausen (1974: 320), IEW (2005: 1011-1012), Kluge/Seebold (2002: 875), Kroonen
(2013: 474), Orel (2003: 371-372), Philippa et al. (2003: 265), Seebold (1970: 462-463).

2.1.7. I-stem verbal abstracts corresponding to class I-VII reduplicated strong verbs

69 PG *fal(l)i- m. ‘fall crash’. Represented in OE fiell, fyll m. ‘id.’ OFris. fal, fel m. ‘id.’
(erth-)fell m. ‘fall to the ground’. Identical radical ablaut grade in e.g. PG *falla- m. ‘crash
fall; end ruin’ > OS fal, fall m. ‘id.’ OHG fal, fall m. ‘id.’ and PG *falla- n. ‘fall’ > ON
fall n. ‘id.’ OE feall n. ‘id.’ OHG (ur-)fal n. ‘breakdown’. Derivationally matching the
class III reduplicated strong verb PG *fallana- s.v. ‘fall’ > ON falla s.v. ‘fall fall in battle’
OE feallan s.v. ‘fall crash’ OFris. falla s.v. ‘fall lower go down collapse’ (conjugated as
a class VI unreduplicated strong verb), OS fallan s.v. ‘fall crash perish’ OHG fallan s.v.
‘fall crash die’. Further etymology uncertain. The traditional view as presented in e.g.
IEW (2005: 851) claims that PG *fal(l)i- and *fallana- should be compared to Arm.
pclanim v. ‘fall’ (for a discussion of the pc in the Armenian forms cf. Klingenschmitt
(1982: 164-172)), pcowl- sb. ‘fall’ and Lith. pùlti v. ‘fall’ (pres. púolu); the two latter forms
developed from PIE *p(h) l-. Unless such a form is to be understood as PIE *peh2-l- with
zero grade PIE *p(h) 2-l-, this reconstruction cannot be maintained in the light of the
present knowledge on the Proto-Indo-European root structure and inventory of phonemes.
A more promising etymology prevails, viz. the one suggested by Praust (2005) that PG
*fallana- is actually a prefigated verb consisting of PIE *h2po prefix ‘off’ and a thematic
present of the root PIE *h3elh1- ‘fall’ as found also in Hitt. hallanai v. ‘trample down
flatten’ Gr. ὄλλυμι v. ‘destroy’ cf. also Scheungraber (2012: 1 with lit.). The combination
of PIE *h2po and the root PIE *h3elh1- is not restricted to Germanic; similar formations
prevail in Gr. ἀπ-όλλυμι v. ‘wreck destroy lose’ (mid. ‘go to waste be ruined be lost’ and
Lat. ab-ole v. ‘destroy banish abolish’ as well as in the Armenian and Lithuanian forms
mentioned above. Literature: Bammesberger (1990: 134), Bjorvand & Lindeman (2000:
199-201), Boutkan & Siebinga (2005: 99-100), Holthausen (1974: 99, 103), IEW (2005:
851), Kluge/Seebold (2002: 273), Kroonen (2013: 125-126), LIV (2001: 463-464), Lloyd

86
& Lühr (2007: 36-40), Orel (2003: 91), Philippa et al. (2009: 461-463), Seebold (1970:
181-182), Sehrt (1966: 114-115), De Vries (1962: 110).
70 PG *fangi- m. ‘catch’. Represented in Goth. ON fengr m. ‘id.’ OE feng m. ‘grasp span
hug’ OFris. fengi, fang, fong m. ‘catch capture’ OHG fang m. ‘id.’. Derivationally
matching the class III reduplicated strong verb PG *fanhana- > *f ana- s.v. ‘catch seize’
> Goth. fahan s.v. ‘catch capture’ ON fá s.v. ‘fetch catch’ OE f n s.v. ‘grasp catch
seize’ OFris. f s.v. ‘id.’ OS f an s.v. ‘id.’ OHG f an s.v. ‘grasp catch seize get’. As
with PG *fal(l)i- and *fallana-, prefigation with PIE *h2po prefix ‘off’ offers an attractive
etymology, the verbal root being either PIE *h1enk̂- ‘obtain’ or as suggested by
Scheungraber (2012: 4-5), PIE *h2enk̂- ‘reach’ (originally *h2nenk̂-). Equally attractive is
the etymology most often suggested, viz. that PG *fanhana- and *fangi- derive from a
nasal present PIE * 2-n-k̂- to the root PIE *peh2k̂- ‘fix fasten’ also found in e.g. Skt.
śa- m. ‘chain’ OLat. pacunt v. ‘agree on fixed terms’. Any attempt of comparison of PG
*fanhana- and *fangi- with forms from the nearly homophonous and originally probably
identical root PIE *peh2ĝ- ‘be fixed’ continued by e.g. Skt. aje v. ‘remains stands by’
Gr. πήγνυμι v. ‘fasten fix moor’ Lat. ang v. ‘id.’ must be abandoned however.
Literature: Bammesberger (1990: 134), Bjorvand & Lindeman (2000: 273-274), Boutkan
& Siebinga (2005: 98, 101), Holthausen (1974: 98, 112), IEW (2005: 787-788),
Kluge/Seebold (2002: 275), Kroonen (2013: 127-128), Lehmann (1986: 102), LIV (2001:
461-462), Lloyd et al. (1998: 9-14, 46), Orel (2003: 92), Philippa et al. (2009: 465),
Seebold (1970: 185-186), Sehrt (1966: 111-114), De Vries (1962: 108, 117).
71 PG * r - m. ‘shout’. Reconstructed as an i-stem by Bammesberger (1990: 136), but as
pointed out by Hinderling (1967: 159), all nominal forms related to the class VI
reduplicated strong verb PG * r ana- s.v. ‘shout’ are reconstructable as a-stems.
Consequently, PG * r - m. ‘shout’ is non-existing. Literature: Bammesberger (1990:
136), Hinderling (1967: 159).
72 PG *stauti- m. ‘thrust push blow’. Represented in ON steytr m. ‘id.’ OFris. st t m.
‘thrust’, OHG st z m. ‘id.’. Derivationally matching the class II reduplicated strong verb
PG *stautana- s.v. ‘hit thrust knock’ > Goth. stautan s.v. ‘hit strike’ OFris. st ta s.v.
‘punch kick’, OS st tan s.v. ‘hit strike’ OHG st zan s.v. ‘push knock; fall; touch’. To be
reconstructed as PIE * s to d-í- to the root PIE * s te d- ‘push prick knock’ cf. also e.g.
Skt. tudáti v. ‘thrusts pushes’ Arm. tcndam v. ‘am shaken’ Alb. shtynj v. ‘shove push’ (<

87
PIE *st d-n- e-), Lat. t nd v. ‘id.’ OIr. (do-)tuit v. ‘falls’ and maybe Lat. st de v. ‘am
eager, am seriously devoted to strive’. Literature: Bammesberger (1990: 133), Bjorvand &
Lindeman (2000: 871-872), Boutkan & Siebinga (2005: 374), IEW (2005: 1032-1034),
Kluge/Seebold (2002: 888), Kroonen (2013: 476), Lehmann (1986: 324), LIV (2001: 601),
Orel (2003: 373), Philippa et al. (2009: 296), Seebold (1970: 463-464), De Vries (1962:
547).
73 PG *walli- m. ‘source spring well’. Represented in OE wiell, wyll m. ‘id.’. Radical o-
grade also in e.g. PG *walla- n. ‘bubbling boiling’ > ON vall n. ‘id.’ OE weall n. ‘id.’.
Derivationally matching the class III reduplicated strong verb PG *wallana- s.v. ‘wave
roar bubble boil’ > OE weallan s.v. ‘id.’ OFris. walla s.v. ‘id.’ OS wallan s.v. ‘id.’
OHG wallan s.v. ‘id.’. Generally seen as related to the root PIE * elh1- ‘turn’ cf. also e.g.
Skt. ūr í- m. ‘wave’ (< PIE * lH-mí-), Lith. vilnìs f. ‘id.’ (< PIE * l 1-ni-), of which it
constitutes an o-grade nasal present, i.e. PIE * ol 1-n-. Given the validity of that analysis,
PG *walli- does not belong to the group of primary i-stems and is thus of no relevance to
the present study. The alternative identification of PG *walli- as a true primary i-stem
derived from PIE * ol 1-í- also cannot be ruled out if Lühr (1976: 76-77) is right in her
analysis of this root and in her general claim that PIE *-RH- > PG *-RR- at least under
certain conditions. Literature: Bammesberger (1990: 134), Bjorvand & Lindeman (2000:
1033-1034), Holthausen (1974: 386, 393), IEW (2005: 1140-1144, esp. 1142),
Kluge/Seebold (2002: 970), Kroonen (2013: 571), Lehmann (1986: 411), LIV (2001: 677),
Orel (2003: 444), Seebold (1970: 538, 552-553), Sehrt (1966: 636-637), De Vries (1962:
641).

2.2. Masculine i-stem verbal abstracts corresponding to a strong verb but displaying unexpected
radical ablaut grade

As against the numerous examples treated above (section 2.1) of masculine i-stem verbal abstracts
derivationally matching the ablaut grade of the stem of the preterite participle of a corresponding
strong verb, we encounter a far smaller quantity of, again mainly masculine, i-stem verbal abstracts
that have the appurtenance to a strong verb in common with the former group. This second group,
however, differs from the regularly formed i-stem verbal abstracts in that the ablaut grades of its
members do not match the stem of the preterite participle.
Examples of such i-stems are listed on the following pages and are, as was also the case with the
former group, categorised in accordance with the class of the strong verb in question.

88
2.2.1. I-stem verbal abstracts corresponding to class I strong verbs

74 No examples.

2.2.2. I-stem verbal abstracts corresponding to class II strong verbs

75 PG *hlauti- m. ‘share lot’. Represented in Goth. hlauts m. ‘lot inheritance’ OE l t, līet

m. ‘share lot; sacrifice’. Radical o-grade also in e.g. PG * la t - f. ‘sacrificial blood; lot’
> ON hlaut f. ‘id.’ and *hlauta- m. ‘share lot’ > ON hlautr m. ‘share lot’ (hapax) OS l t
m. ‘id.’ OHG l z, l z m./n. ‘id.’. Etymologically belonging to the class II strong verb PG
*hleutana- s.v. ‘obtain by lot’ but its ablaut grade is aberrant. If created to the weak stem
of the verb, i.e. the stem found in the preterite participle, the i-stem should have been PG
*hluti- m. ‘lot’ which actually exists as a competing form. Literature: Bammesberger
(1990: 133), Casaretto (2004: 180), IEW (2005: 604-605, esp. 605), Kluge/Seebold (2002:
582), Kroonen (2013: 230, 233), LIV (2001: 105-106, 365), Orel (2003: 175), Seebold
(1970: 264-265), Sehrt (1966: 262), De Vries (1962: 235).
76 PG *laudi- m. ‘form appearance’. Represented in Goth. (jugga-)lauþs m. ‘young man
youth’. Etymologically belonging to the class II strong verb PG *leudana- s.v. ‘grow’ >
Goth. liudan s.v. ‘id.’ ON loðinn adj./ptc. ‘shaggy dishevelled’ OE l odan s.v. ‘grow’
OS liodan s.v. ‘id.’ OHG liotan s.v. ‘id.’ but its ablaut grade is aberrant. Descendants of
the expected form PG *ludi- are not attested. Attested are, on the other hand, descendants
of another i-stem to this root, viz. PG *leudi- with radical full grade. The i-stem shape of
Goth. jugga-lauþs (-laudi-) stands a good chance of being secondary seeing that Goth. -
laudi- as well as the more productive Goth. -lauda- in e.g. swa-lauþs adj. ‘so big’ for that
matter, only appear as second members of compounds. In that position, when the second
member of the compound is adjectival, i.e. when the compound is exocentric, transition of
thematic stems (a-stems) into i-stems is, from a Proto-Indo-European point of view, an
expected and regular process, examples of which include Skt. gandhá- m. ‘smell
fragrance’  d ū á-gandhi- adj. ‘smelling of smoke having the smell of smoke’ Gr.
μισθός m. ‘wages pay hire’  ἀ-μισθί- adj. ‘without pay’ and as is the case here Goth.
*lauda- m. ‘shape appearance’  jugga-laudi- m. ‘having a young appearance’ cf.
further e.g. Brugmann & Delbrück (1906: 112-113) and Rasmussen (1988 [1999]: 320).
The i-stem, or rather the underlying thematic a-stem, is to be reconstructed as PIE
*h1l dh-o- to the root PIE *h1le dh- ‘rise grow’ also known from e.g. Skt. vī-rúdh- f.
‘plant’ (< PIE * -h1lúdh-), ródhati, róhati v. ‘grows’ Gr. ἤλυθον v. ‘came’ (aor. < PIE

89
*é-h1ludh-e-), ἐλεύσομαι ‘will come’ (fut. < PIE *h1l dh-(h1)s-e-), OIr. luid v. ‘went’.
Literature: Bammesberger (1990: 133), Bjorvand & Lindeman (2000: 540), Casaretto
(2004: 180), Holthausen (1974: 199), IEW (2005: 306-307, 684-685), Kluge/Seebold
(2002: 579), Kroonen (2013: 332), Lehmann (1986: 234-235), LIV (2001: 248-249), Orel
(2003: 242), Philippa et al. (2007: 224), Seebold (1970: 335-336), Sehrt (1966: 340), De
Vries (1962: 363).
77 PG *leudi- m. ‘man’ (pl. ‘people’) as represented in ON ljóðr, lýðr m. ‘id.’ OE l od m.
‘man king’ f. ‘people’ OFris. liude m. ‘id.’ OS liodi m. ‘id.’ OHG liut m. ‘id.’
constitutes a second aberrant i-stem form to PG *leudana- s.v. ‘grow’. Further cognates
with full grade of the root include Lith. liáudis f. ‘people’ OCS lj dъ m. ‘id.’ pl. lj dьje),
and for the special semantic development of ‘grow’ > ‘growth-group’ > ‘(free) stock
people’ cf. Gr. ἐλεύθερος adj. ‘free’ and Lat. *lūber > līber adj. ‘free’ m. ‘god of
vegetation and growth’; both < PIE *h1l dh-ero-. The creation of a PG *leudi- next to the
likewise secondary form PG *laudi- ‘form appearance, is difficult to motivate. Neither is
PG *leudana- a class V strong verb in which case radical full grade would be expected;
nor does any a- or s-stem with radical full grade, i.e. PG *leuda- < PIE *h1le dh-o- or PG
*leudaz ~ *leudiz- < PIE *h1l dh-os ~ *h1l dh-es- which could otherwise have served as
a source of analogy for the creation of PG *leudi-, prevail in any of the Indo-European
branches. In my view, either of the explanations presented by Casaretto (2004: 167), viz.
that it has been analogically influenced by i-stem verbal abstracts corresponding to class V
strong verbs, and Hinderling (1967: 40), viz. that the phonological identity of the present
stem and the preterite participle stem in the class V and VI strong verbs have caused the
derivatory basis to be reanalysed as the present stem, are by far the most convincing ones,
at least for want of better and more plausible alternatives. Literature: Bammesberger
(1990: 136), Bjorvand & Lindeman (2000: 555-556), Boutkan & Siebinga (2005: 242),
Holthausen (1974: 199), IEW (2005: 306-307, 684-685), Kluge/Seebold (2002: 572),
Kroonen (2013: 332-333), LIV (2001: 248-249), Orel (2003: 242), ), Philippa et al. (2007:
224), Seebold (1970: 335), Sehrt (1966: 345-346), De Vries (1962: 359-360, 369).
78 PG *rauki- m. ‘smoke’. Represented in ON reykr m. ‘id.’ OE riec, r c m. ‘id.’ OFris. r k
m. ‘id.’ OS r k m. ‘id.’ OHG rouh m. ‘id.’. Etymologically belonging to the class II
strong verb PG *reukana- s.v. ‘smoke’ (originally ‘smell’), but its ablaut grade is aberrant.
This i-stem with radical o-grade coexists with the expected zero grade form PG *ruki- m.

90
‘smell; smoke’. The solution to the problem which of these two i-stems is more original
may profitably be sought in the verb PG *raukiana- w.v. I ‘smoke’ which is often seen
analysed as a denominal verb with PG *rauki- as its derivatory base. If, however, rather
analysed as a causative i.e. ‘make smell’ > ‘smoke’ PG *raukiana- can easily serve if not
as a derivatory base for an i-stem PG *rauki- m. ‘smoke’ then at least as a heavy source of
inspiration for a separation of the original PG *ruki- m. ‘smell; smoke’ into *ruki- ‘smell’
on the basis of the strong verb PG *reukana- s.v. ‘smoke’ (originally ‘smell’) and newly
shaped *rauki- m. ‘smoke’ on the basis of or analogically influenced by the causative PG
*raukiana- w.v. I ‘smoke’. Literature: Bammesberger (1990: 133), Bjorvand & Lindeman
(2000: 742-743, 747-748), Holthausen (1974: 257, 259), IEW (2005: 871-872),
Kluge/Seebold (2002: 746, 764-765), Kroonen (2013: 406), Orel (2003: 299), Philippa et
al. (2003: 681-682, 689-690), Seebold (1970: 379-380), Sehrt (1966: 440), De Vries (1962:
443, 449).
79 PG *saudi- m. ‘meat broth’. Represented in Goth. sauþs m. ‘sacrifice’ ON sauðr m.
‘sheep’ (with unumlauted vowel; the umlauted variant is found in OGutn. soyþr m.
‘sheep’) the semantical development being one of ‘meat broth’ > ‘boiled meat’ >
‘sacrifice’ and ‘sheep’ (< ‘raw meat for cooking’) respectively cf. Bjorvand & Lindeman
(2000: 762) and Seebold (1970: 401). Radical o-grade also in e.g. PG *sauþa- m. ‘broth’ >
OGutn. sauþr m. ‘well’ OE s at m. ‘well hollow pit lake’ OFris. s t m. ‘well’ and
PG *sauda- m. ‘cooking boiling’ > Norw. saud m. ‘cooking’ MHG s t m. ‘boiling
cooking; well’. Etymologically belonging to the class II strong verb PG *seuþana- s.v.
‘boil seethe’, but the attested o-grade is unexpected in the light of the numerous other
Germanic i-stem verbal abstracts treated in section 2.1. PG *saudi- with radical o-grade (<
PIE *h2so t-í- if inherited from Proto-Indo-European) coexists with the expected zero
grade form PG *sudi- m. ‘decoction extract’. Two explanations for this coexistence come
to mind. Either PG *saudi-, having received its o-grade due to influence from either of the
a-stems PG *sauþa- m. or *sauda- m. mentioned above, is a recent formation in
comparison with PG *sudi-, or PG *saudi- is simply to be regarded as the original i-stem
form inherited from Proto-Indo-European in a period when the Proto-Germanic process of
creating an i-stem verbal abstract to the weak stem of the verb, i.e. the stem of the preterite
participle, had yet to become automatised. Literature: Bammesberger (1990: 133),
Bjorvand & Lindeman (2000: 762-763), Casaretto (2004: 182), Holthausen (1974: 287),

91
IEW (2005: 914-915), Kroonen (2013: 428), LIV (2001: 285), Orel (2003: 320), Seebold
(1970: 400-401), De Vries (1962: 464).
80 PG *smauki- m. ‘smoke’. Represented in OE s īc, s īec m. ‘id.’. Etymologically
belonging to the class II strong verb PG *smeukana- s.v. ‘smoke’ > OE s ocan s.v. ‘id.’
but as with the previous i-stem formations of the present paragraph, radical o-grade is
unexpected, the only other formation with radical o-grade being the weak verb PG
*smaukiana- w.v. I ‘smoke’ > OE s īcan, s īecan s.v. I ‘id.’. At first sight PG *smauki-
and *smeukana- descend from a form with a root-final PIE *g. When including forms with
relevant semantics from outside Germanic, however, we are faced with a multitude of root-
final velars: PIE *k in Arm. mowx sb. ‘smoke’ (< PIE *s o kHo- or maybe rather PIE
*s o H-ko-; for the development PIE *-h1/2-t- > *-th- cf. Olsen (1994: 274-275)) and OIr.
múch f. ‘fire’ (< PIE *smuHk-), PIE *k or *gh in Gr. σμύχω v. ‘burn up smolder away’ (<
PIE *smuHk/gh-) and Lith. smáugti v. ‘choke’ (< PIE *s o Hgh- e-), and finally PIE *g in
the Germanic forms at least on the surface. The effects of Kluge’s Law (PIE *-K-n- > PG
*-kk-), which may have been in operation in the iterative PG *s kk ja na- w.v. II
‘smoke’ as continued in OE smocian s.v. II ‘id.’ and G(Lux.) schmocken s.v. II ‘id.’ reveal
that the PG root *smeuk- need not mirror a PIE *s e H g- but can actually mirror also
*s e H)k- and *s e H gh-. Literature: Bammesberger (1990: 133), Holthausen (1974:
301-303), IEW (2005: 971), Kroonen (2013: 458-460), Orel (2003: 354), Seebold (1970:
440-441).

2.2.3. I-stem verbal abstracts corresponding to class III strong verbs

81 PG *balgi- m- ‘sack bag’. Represented in Goth. balgs m. ‘skin-bag’ ON belgr m. ‘flayed

animal skin bag; bellows; belly’ OE belg, bylig, bielg m. ‘sack bag; bellows’ OFris.
bl s-)balch m. ‘bellows’ OHG balg m. ‘skin tube pod’. Etymologically belonging to the
class III strong verb PG *belgana- s.v. ‘swell’ > ON bolginn adj./ptc. ‘swollen’ OE belgan
s.v. ‘become angry be angered’ OFris. (over-)bulgen adj./ptc. ‘resentful angry incensed’
OS belgan s.v. ‘become angry be angered’ OHG belgan s.v. ‘id.’ the expected i-stem
verbal abstract of which would be PG †bulgi- rather than *balgi-. To be reconstructed as
PIE *bholĝh-í- to the root PIE *bhelĝh- ‘swell’ also found in e.g. Skt. bar í - n. ‘sacrificial
straw’ Av. barәz - n. ‘pillow’ OIr. bolg m. ‘sack’ f. ‘bladder’ (< PIE *bholĝh-o-),
bolgaim v. ‘swell’. Literature: Bammesberger (1990: 133), Bjorvand & Lindeman (2000:
67-69), Boutkan & Siebinga (2005: 36, 301), Casaretto (2004: 178-179), Holthausen

92
(1974: 19, 22), IEW (2005: 125-126), Kluge/Seebold (2002: 85), Kroonen (2013: 49, 58),
Lehmann (1986: 59-60), LIV (2001: 73-74), Lloyd & Springer (1988: 438-440, 528-530),
Orel (2003: 33-34, 41), Philippa et al. (2003: 228, 261), Seebold (1970: 99-101), Sehrt
(1966: 43), De Vries (1962: 32, 49).
82 PG *bandi- m. ‘captivity’. Represented in OE bend m./f. ‘id.’ OFris. bend m. ‘id.’.
Radical o-grade also in e.g. PG *banda- n. ‘string band bond cord chain’ > ON band n.
‘id.’ OFris. band n. ‘id.’ OS b d-)band n. ‘crown’ OHG bant n. ‘prison; band priest’s
band’. Etymologically belonging to the class III strong verb PG *bindana- s.v. ‘bind
chain’ > Goth. bindan s.v. ‘id.’ ON binda s.v. ‘bind’ OE bindan s.v. ‘id.’ OFris. binda
s.v. ‘bind chain’ OS bindan s.v. ‘id.’ OHG bintan s.v. ‘bind chain capture’. To be
reconstructed as PIE *bhondh-í- to the root PIE *bhendh- ‘bind’ with numerous cognates
outside Germanic, e.g. Skt. bad n t v. ‘binds’ bandhá- m. ‘bond fetter’ Av. banda- m.
‘id.’ basta- adj./ptc. ‘bound’ (< PIE *bhndh-tó-), OPers. basta- adj./ptc. ‘id.’ Gr. πεῖσμα n.
‘rope on a ship cable’ (< PIE *bhendh-s -n), Lat. (of-)fendix f. ‘chin strap knot or band to
keep the headwear of the priest in position’ OIr. buinne f. ‘bond bracelet’ (< PIE *bhondh-
e 2-), Lith. be dras m. ‘companion participant’ bandà f. ‘cattle’ and maybe Gr. πάσχω v.
‘get an impression experience; tolerate endure’ (< PIE *bhndh-sk̂ -) ἔπαθον v. ‘id.’ (aor.
< PIE é-bhndh-e-) if we can assume a semantic development of ‘be bound’ > ‘be tense’.
Literature: Bammesberger (1990: 134), Bjorvand & Lindeman (2000: 74-75), Boutkan &
Siebinga (2005: 37, 49), Holthausen (1974: 19, 23), IEW (2005: 127), Kluge/Seebold
(2002: 87, 124), Kroonen (2013: 51, 64), Lehmann (1986: 71), LIV (2001: 75), Lloyd &
Springer (1988: 462-463), Lloyd et al. (1998: 72-75), Orel (2003: 35, 41), Philippa et al.
(2003: 215, 316-317), Seebold (1970: 102-104), Sehrt (1966: 51-52, 265), De Vries (1962:
25, 37).
83 PG *dranki- m. ‘drink’. Represented in OE drenc m. ‘id.’. Radical o-grade also in e.g. PG
*dranka- m. ‘drink’ > OFris. drånk m. ‘id.’ OS drank m. ‘id.’ OHG tranc m. ‘id.’ and PG
*dranka- n. ‘drink’ > Goth. dragk n. ‘id.’ OHG tranc n. ‘id.’. Etymologically belonging to
the class III strong verb PG *drinkana- s.v. ‘drink’ whose expected i-stem verbal abstract
is also attested as reflects of PG *drunki- m. ‘drink’. Seeing that both PG *drunki-,
*dranki- and *drinkana- mirror a nasal present to a root PIE *dhreĝ- ‘glide move’ and not
a root itself, both i-stems stand virtually no chance of being an archaism, and with PG
*drunki- representing the synchronically expected form, an alternative explanation must be

93
sought for PG *dranki-. The most obvious one is that it has been influenced by or
secondarily reshaped from the o-grade a-stem. Literature: Casaretto (2004: 78), Holthausen
(1974: 77, 79), IEW (2005: 273), Kluge/Seebold (2002: 924), Kroonen (2013: 103),
Lehmann (1986: 94-95), Orel (2003: 74), Philippa et al. (2003: 623), Seebold (1970: 165-
166), Sehrt (1966: 83).
84 PG *stangi- m. ‘bar pole staff stake’. Represented in OE steng m. ‘id.’. Radical o-grade
also in e.g. PG *stang - f. ‘bar pole staff stake’ > ON stǫng f. ‘id.’ (partially inflected as
a root noun, cf. Ph.d. article no. 1), OS stanga f. ‘id.’ OHG stanga f. ‘id.’. Etymologically
belonging to the class III strong verb PG *stingana- s.v. ‘stick thrust’ whose expected i-
stem verbal abstract should appear as PG *stungi-; a reconstructed form to which cognates
are actually attested. Two questions now arise, viz. what could have motivated the
coexistence of these two i-stem verbal abstracts, and which form is archaic. The most
straightforward solution to the problem would be to assume that PG *stangi- < PIE
*stongh-í- is the more archaic form since the formation of it cannot be motivated directly.
PG *stungi-, on the other hand, is easily explicable as productively derived from the ablaut
grade found in the stem of the preterite participle of the corresponding strong verb. It is
also possible, however, that *stangi- is the unoriginal form. In that case, we might consider
invoking influence from PG *stang - f. ‘bar pole staff stake’ on PG *stangi- m. ‘id.’.
Finally, though disinclined to believe in that explanation himself, Seebold (1970: 462)
mentions the theoretical possibility of the hapax Goth. (us-)stagg s.v. ‘pluck out’ (ipv.)
vouching for the existence of a class III reduplicated strong verb PG *stangana- ‘sting
stick’ s.v. competing with *stingana- s.v. ‘id.’. Needless to say the synchronically
expected form of an i-stem verbal abstract to PG *stangana- would be PG *stangi-.
Literature: Bammesberger (1990: 134), Holthausen (1974: 319), IEW (2005: 1014-1015),
Kluge/Seebold (2002: 875), Kroonen (2013: 474), LIV (2001: 589), Orel (2003: 371),
Seebold (1970: 462, 470-471), De Vries (1962: 559).
85 PG *stankwi- m. ‘smell’. Represented in OE stenc m. ‘odour; smell’. Radical o-grade also
in e.g. PG *stankwa- m. ‘clash; smell’ > ON stǫkkr m. ‘sudden movement’ OE stanc m.
‘spraying sprinkling’ OS stank m. ‘odour pleasant smell’ OHG stanc m. ‘odour; smell’.
Etymologically belonging to the class III strong verb PG *stinkwana- s.v. ‘thrust clash;
stink’. As was also the case with the previous lemma, i.e. PG *stangi- m. ‘bar pole staff
stake’ this aberrant form coexists with the structurally expected i-stem verbal abstract PG

94
*stunkwi- m. ‘smell’. Here too an estimate is wanted as to what form is original and
again two major options present themselves, viz. of the o-grade form to be original due to
the zero grade form being easily constructible from PG *stinkwana- or of the o-grade form
to have been remodelled from PG *stunkwi- in analogy with the a-stem PG *stankwa-.
Literature: Bammesberger (1990: 134), Holthausen (1974: 317, 319), IEW (2005: 1033),
LIV (2001: 596-597), Orel (2003: 371), Philippa et al. (2009: 264-265), Seebold (1970:
471-472), Sehrt (1966: 504-505), De Vries (1962: 559).
86 PG *swalgi- m. ‘abyss; swirl whirlpool’. Represented in ON svelgr m. ‘swirl whirlpool’
OE svelg m. ‘abyss’. Etymologically belonging to the class III strong verb PG *swelgana-
s.v. ‘swallow’ to which also another i-stem verbal abstract PG *swulgi- m. ‘drink
swallow gulp’ is related. For the question of which form is the more archaic one cf. the
discussion under the previous two lemmata. It is important to note, however, that in the
case of PG *swalgi-, no corresponding a- or ō-stem exists that could have otherwise
facilitated any transition from either of the thematic paradigms to the i-stem paradigm.
Literature: Bammesberger (1990: 134), Bjorvand & Lindeman (2000: 882), IEW (2005:
901, 1145), LIV (2001: 289-290, 530-531, though no mentioning of the Germanic forms),
Orel (2003: 390), Seebold (1970: 488-489), De Vries (1962: 334).
87 PG *swangwi- m. ‘swing; stroke’. Represented in OE sweng ‘id.’ OFris. sweng m.
‘perfusio i.e. the act of pouring water over a patient affusion moistening’. Etymologically
belonging to the class III strong verb PG *swingwana- s.v. ‘swing’ > OE swingan s.v.
‘swing flee; strike discipline castigate’ OFris. swinga s.v. ‘pour over douse’ OS
swungan s.v. ‘swung fallen’ (pret.ptc.), OHG swingan s.v. ‘swing’ the i-stem verbal
abstract of which one would have rather expected to find as PG †swungwi-. Further
etymology highly uncertain. At least, the classic comparandum mentioned by e.g. IEW
(2005: 1047-1048) of PG *swengwana- and Skt. svájate v. ‘hugs embraces’ must be
abandoned on formal grounds (Skt. j < PIE *g(w); PG *gw < PIE *gwh or *kw). Literature:
Bammesberger (1990: 134), Boutkan & Siebinga (2005: 383-384), Holthausen (1974: 335,
338), IEW (2005: 1047-1048), Kluge/Seebold (2002: 831, 835), LIV (2001: 611), Orel
(2003: 394), Philippa et al. (2003: 681), Seebold (1970: 493), Sehrt (1966: 521).
88 PG *swanki- m. ‘misery sorrow; toil; temptation’. Represented in OE swenc m. ‘id.’.
Radical o-grade also in e.g. PG *swank - f. ‘swing’ > Norw. (dial.) svokk f. ‘arch of the
foot’. Etymologically belonging to the class III strong verb PG *swinkana- s.v. ‘toil

95
labour’ > OE swincan s.v. ‘id.’. Extra-Germanic comparanda are limited, cf. e.g. OIr. seng
adj. ‘slender slim’ (< PIE *s eng-o-) and maybe Skt. svájate v. ‘hugs embraces’ which
actually stands a better chance on comparison with PG *swanki- than with *swangwi-.
Literature: Bammesberger (1990: 134), Holthausen (1974: 335, 338), IEW (2005: 1047-
1048, esp. 1047), LIV (2001: 610-611), Orel (2003: 391), Seebold (1970: 493-494).
89 PG *sweli- m. ‘callosity callous skin’. Represented in OS swil n. ‘id.’ OHG swil n. ‘id.’.
Etymologically belonging to the class III strong verb PG *swellana- s.v. ‘swell’ whose
synchronically regular i-stem verbal abstract appears as PG *swuli- m. ‘swelling’. In the
light of the neuter gender of the attested forms, Seebold (1970: 490) suggests that PG
*sweli- has transcended from the inflectional class of the s-stems to that of the i-stems. By
means of reanalysis of the weak s-stem suffix PIE *-es- > PG *-iz- as PIE *-i-z,
transformation of s-stems into i-stems has become a trivial process in Germanic, or at least
in Gothic and West Germanic, as outlined by e.g. Van Helten (1910-1911: 500-504),
Hinderling (1967: 102-116) and Bammesberger (1990: 137-138). Literature:
Bammesberger (1990: 136), Kluge/Seebold (2002: 835), Kroonen (2013: 499), LIV (2001:
609-610), Orel (2003: 394), Seebold (1970: 489-490).
90 PG *weni- m. ‘friend’. Represented in ON vinr m. ‘friend’ OE wine m. ‘friend protector’
OS wini m. ‘friend’ OHG wini m. ‘id.’. Etymologically belonging to the class III strong
verb PG *winnana- s.v. ‘exert oneself take pains’ whose synchronically regular i-stem
verbal abstract appears as PG *wunni- m. ‘suffering pain’. In the light of the extra-
Germanic s-stems Skt. vánas- n. ‘desire’ and Lat. venus f. ‘(goddess of) love’ reanalysis of
this s-stem PIE * énH-os ~ * énH-es- as an i-stem constitutes an obvious explanation of
this aberrant i-stem, cf. PG *sweli- m. ‘callosity callous skin’for a similar analysis.
Literature: Bammesberger (1990: 136), Bjorvand & Lindeman (2000: 1035), Holthausen
(1974: 397), IEW (2005: 1146-1147), Kroonen (2013: 579), LIV (2001: 682-683), Orel
(2003: 455), Seebold (1970: 556-557), De Vries (1962: 666).
91 PG *w n - m./f. ‘hope expectation’. Represented in Goth. wens f. ‘id.’ ON ván, vón, ón f.
‘id.’ (early transition into the ō-stem declension), OE w n f. ‘supposition opinion’ OFris.
w n f. ‘idea opinion’ OS w n m. ‘hope expectation’ OHG w n m. ‘hope expectation
opinion elusion’. Along with the previous lemma PG *w n - belongs etymologically to
the class III strong verb PG *winnana- s.v. ‘exert oneself take pains’ with the
synchronically regular PG *wunni- m. ‘suffering pain’. The obvious explanation for the

96
radical lengthened grade in PG *w n - is vrddhi, maybe under analogical influence from
the lengthened grade i-stem adjectives, cf. section 2.4. Literature: Bammesberger (1990:
136), Bjorvand & Lindeman (2000: 1069-1070), Casaretto (2004: 187), Holthausen (1974:
404), IEW (2005: 1146-1147), Kluge/Seebold (2002: 968), Kroonen (2013: 579),
Lehmann (1986: 401), LIV (2001: 682-683), Orel (2003: 461), Philippa et al. (2009: 580),
Seebold (1970: 556-557), Sehrt (1966: 638), De Vries (1962: 643).

2.2.4. I-stem verbal abstracts corresponding to class IV strong verbs

92 PG *drapi- m. ‘blow stroke’. Represented in OE drepe m. ‘id.’. Etymologically belonging

to the class IV strong verb PG *drepana- s.v. ‘blow strike’ and coexisting with the
synchronically regular i-stem abstract PG *drupi- m. ‘blow stroke’. Literature:
Bammesberger (1990: 135), Holthausen (1974: 77), IEW (2005: 272-273), LIV (2001:
153), Orel (2003: 75), Seebold (1970: 166-167).
93 PG *kwemi- m. ‘eventus i.e. outcome result’. Represented in OHG (ga-)quimi m. ‘id.’.
Etymologically belonging to the class IV strong verb PG *kwemana- s.v. ‘come’ and
coexisting with PG *kwumi- m. ‘coming’, which must be regarded as the structurally
expected form. Since no other full grade nominal derivatives of PG *kwemana- exist
except for PG *kweman- m. ‘newcomer’ > OHG (niuwi-)qhuemo m. ‘novice’ no other
forms could have triggered or facilitated the creation of PG *kwemi-. In my view, either of
the explanations suggested for PG *leudi-, must therefore be regarded as the only option.
Literature: Bammesberger (1990: 136), IEW (2005: 464-465), Kroonen (2013: 316), LIV
(2001: 209-210), Orel (2003: 227-228), Seebold (1970: 315-317).

2.2.5. I-stem verbal abstracts corresponding to class V strong verbs

94 PG *mati- m. ‘food’. Represented in Goth. mats m. ‘id.’ ON matr m. ‘id.’ OE mete m.

‘id.’ OFris. mete m. ‘id.’ OS meti m. ‘id.’. Radical o-grade also in e.g. PG *mata- n.
‘food’ > OS mat n. ‘id.’ OHG maz m. ‘id.’. Etymologically belonging to the class V strong
verb PG *metana- s.v. ‘measure’ > Goth. mitan s.v. ‘id.’ ON meta s.v. ‘id.’ OE metan s.v.
‘id.’ OFris. meta s.v. ‘id.’ OS metan s.v. ‘regard’ OHG mezzan s.v. ‘measure’ the
semantic development of PG *mati- being one of ‘measuring’ > ‘part portion’ > ‘food’.
To be reconstructed as PIE *mod-í- to the root PIE *med- ‘measure regard; worry give
advice give healing’ with reflects in e.g. Gr. μή ομαι v. ‘consider decide’ and Lat.
meditor v. ‘consider carefully’ medeor v. ‘heal come to assistance’. This etymology

97
however, is only one of the two etymologies most often adduced for PG *mati-. The
second etymological proposal separates PG *mati- m. ‘food’ from PG *metana- s.v.
‘measure’ only to compare it with the root PIE *mad- (IEW 2005: 694-695) or *med- (LIV
2001: 423-424) ‘wet; glossy fat well-fed’ with numerous descendants in the individual
branches, cf. e.g. Skt. mándati v. ‘rejoices at; enjoys gets drunk’ mádati v. ‘rejoices at;
enjoys gets drunk gets enough to eat’ máda- m. ‘intoxicant’ Lat. ade v. ‘am drunk’.
Literature: Bammesberger (1990: 135), Bjorvand & Lindeman (2000: 591-592), Boutkan
& Siebinga (2005: 253-254, 259), Casaretto (2004: 180-181), Holthausen (1974: 220),
IEW (2005: 694-695, 705-706), Kluge/Seebold (2002: 614-615), Kroonen (2013: 378),
Lehmann (1986: 376-377), LIV (2001: 423), Orel (2003: 263), Seebold (1970: 352-354),
Sehrt (1966: 368, 377), De Vries (1962: 380, 385).
95 PG *staki- m. ‘mark scar’. Represented in Goth. staks m. ‘id.’. Identical radical ablaut
grade in e.g. PG *stakan- m. ‘pole post stake’ > OE staca m. ‘id.’ OFris. staca m. ‘id.’.
Etymologically belonging to the class V strong verb PG *stekana- s.v. ‘stick thrust’ which
may, in turn, be a backformation through reinterpretation of the a-umlauted form PG
*stekana- of an original zero grade verb PG *stikana-, cf. Seebold (1970: 468). The
expected i-stem verbal abstract of this verb is PG *stiki- m. ‘prick stab’. The aberrant form
PG *staki- cannot be explained as archaic seeing that the verb to to which it corresponds is,
in itself, secondary. In my view, the most obvious candidate for the analogy of creating an
i-stem verbal abstract with radical o-grade is the i-stem PG *stangi- m. ‘bar pole staff
stake’ to which semantic parallels can be drawn from PG *stekana-. Literature:
Bammesberger (1990: 135), Casaretto (2004: 182), Holthausen (1974: 314), IEW (2005:
1014, 1016-1017), Kroonen (2013: 472, 476), Lehmann (1986: 322), LIV (2001: 592-593),
Orel (2003: 370), Philippa et al. (2009: 254), Seebold (1970: 467-468).

2.2.6. I-stem verbal abstracts corresponding to class VI strong verbs

96 No examples.

2.2.7. I-stem verbal abstracts corresponding to class I-VII reduplicated strong verbs

97 No examples.

98
2.3. I-stem nouns with no correspondence to a strong verb

Some primary i-stems, of masculine as well as of feminine gender, resist further analysis within
Germanic. They are, in other words, not to be synchronically regarded as verbal abstracts derived
from strong verbs.
On a diachronic basis, however, a decent portion of them should actually be categorised as such
seeing that in those cases the “strong” verbs lacking in Germanic are attested as thematic presents
in other Indo-European branches. This is the case for, e.g., PG *duni- m. ‘noise’ with a perfect
match in Skt. dhvánati v. ‘roars sounds’ and maybe PG *hwali- m. ‘whale’ if it is to be understood
as an agent noun to the root PIE *kwelh1- ‘turn’ i.e. PIE *kwólh1-i- ‘turner’ with a thematic present
attested in e.g. Skt. cárati v. ‘moves walks’ and Gr. πέλομαι, πέλω v. ‘move am located am
become’ and Alb. shell v. ‘bring carry; turn’.
Alphabetically listed examples of such i-stem nouns that, for one reason or another, cannot be
matched with any strong verbs in Germanic are found on the following pages.

98 PG *awi- f. ‘ewe female sheep’. Not reflected directly in any Germanic languages except
perhaps for OE ewe f. ‘id.’; we find it only in compounds cf. e.g. PG *awi-st-ra- m.
‘sheepfold’ > Goth. awistr n.? ‘id.’ OE owestre, westre m./f. ‘id.’ OHG ewist, æwist m.
‘id.’ and as secondarily enlarged into PG *awī- ~ *awj - f. ‘ewe female sheep’ > ON ær f.
‘id.’ OE ow f. ‘female lamb’ OFris. ei f. ‘id.’ OS euwi, ewi f. ‘id.’ OHG ou, ouwi f.
‘sheep’. To be reconstructed as an unanalysable acrostatic i-stem PIE *h2 -i- ~ *h2 -i-
‘sheep’ cf. e.g. Pinault (1997: 190-193), NIL (2008: 335-339) and Cohen & Hyllested
(2012: 65-66), the descendants of which are represented in most branches of Indo-
European, cf. e.g. CLuw. w - c. ‘id.’ HLuw. hawa/i- c. ‘id.’ Skt. ávi- f. ‘id.’ Gr. ὄ ς f.
‘id.’ Lat. ovis f. ‘id.’ Lith. avìs f. ‘id.’ OCS ovьca f. ‘id.’. Literature: Bammesberger
(1990: 127), Holthausen (1974: 93), IEW (2005: 784), Kroonen (2013: 45), Lehmann
(1986: 53), NIL (2008: 335-339), Orel (2003: 31), Philippa et al. (2007: 464), De Vries
(1962: 681).
99 PG *bak(w)i- m. ‘creek brook rivolet’. Represented in ON bekkr m. ‘id.’ (inflected as a ja-
stem), OE bece m./n. ‘id.’ OFris. -bitze m. ‘id.’ OS beki m. ‘id.’ OHG bah m. ‘id.’.
Either to be compared to ORu. bagъno n. ‘mud marsh’ or to be reconstructed as PIE
*bhog(w)-í- to the root PIE *bhegw- ‘run’; for the latter cf. e.g. Gr. φέβομαι v. ‘flee’ φόβος
m. ‘flight escape; fear’, Lith. b gt v. ‘run flee’ b g s m. ‘flight escape; run’ and maybe

99
MIr. búal f. ‘running water’ (< PIE *bhogw-leh2-), búar m. ‘diarrhoea’ (< PIE *bhogw-ro-).
Literature: Bjorvand & Lindeman (2000: 66-67), Holthausen (1974: 18), IEW (2005: 116,
161), Kluge/Seebold (2002: 80), Kroonen (2013: 48), LIV (2001: 67), Lloyd & Springer
(1988: 427-429), Orel (2003: 33), Philippa et al. (2003: 243-244), De Vries (1962: 31).
100 PG *duni- m. ‘noise’. Represented in ON dynr m. ‘id.’ (with y probably from dynja w.v. I
‘roar gush’) OE dyne m. ‘id.’ OHG tuni m. ‘id.’. Serves as the basis for the denominal
verb PG *dunjana- w.v. I ‘make noise boom roar’ > ON dynja w.v. I ‘roar gush’ OE
dynnan w.v. I ‘make a noise resound’ OS dunnian w.v. I ‘roar rumble’ which may
however, also be analysed as a zero grade causative formation. To be reconstructed as PIE
*dhun-í- to the root PIE *dh en- ‘sound’ with extra-Germanic cognates in Indic only, cf.
e.g. Skt. dhvánati v. ‘roars sounds’ dhvaní- m. ‘sound thunder word’ dhúni- adj.
‘roaring sounding’ dhunáyati v. ‘roars’. The i-stem PG *duni- and Skt. dhúni- are exact
cognates with the sole exception that Skt. dhúni- (< PIE *dhún-i-) is the radically stressed
adjective/agent noun and PG *duni- (< PIE *dhun-í-) is likely to be the suffixally stressed
abstract noun formed from the former by means of internal derivation. In Indic, the
abstract noun appears in the shape of a full or o-grade formation Skt. dhvaní-, of which the
latter is only possible on a regular basis if the root contained a final laryngeal. As already
mentioned, the weak verb PG *dunjana- is to be seen either as a denominal verb shaped
from PG *duni-, i.e. PIE *dh n- -, or as a zero grade causative formation, i.e. PIE dh n-
e-, which is probably the better alternative in the light of the exact parallel of Skt.
dhunáyati. Literature: Bammesberger (1990: 132), Holthausen (1974: 81-82), IEW (2005:
277), LIV (2001: 158), Orel (2003: 79), Sehrt (1966: 88), De Vries (1962: 90).
101 PG *fadi- m. ‘lord’. Represented in Goth. -faþs m. ‘lord master’. Extra-Germanic cognates
of this PIE *póti- ‘lord master husband’ abound cf. e.g. Toch. A pats m. ‘husband’ Skt.
páti- m. ‘lord husband’ Gr. πόσις m. ‘husband’ Lith. pàts m. ‘husband married man’
(OLith. patis). Substantivised from the i-stem adjective PIE *póti- adj. ‘capable’ cf. e.g.
OE fad adj. ‘strong brave big’ and Lat. potis adj. ‘capable mighty’ which might in turn
have developed from a particle PIE *pot(i) ‘self’ cf. e.g. Hitt. -pat ‘also even’ Lith. pàt
‘id.’. Literature: Bammesberger (1990: 127), Casaretto (2004: 184), IEW (2005: 842),
Kroonen (2013: 121), Lehmann (1986: 82-83), Orel (2003: 88).
102 PG *hugi- m. ‘intellect mind’. Represented in Goth. hugs m. ‘understanding mind’ ON
hugr m. ‘mind mood’ OE hyge m. ‘mind heart soul’ OFris hei m. ‘mind’ OS hugi m.

100
‘thought mind’ and albeit transferred to the u-stem inflection, OHG hugu m. ‘mind
thought spirit’. Serves as the basis for the denominal verb PG *hugjana- w.v. I ‘think’ >
Goth. hugjan w.v. I ‘id.’ ON hyggja w.v. I ‘think mean believe’ OE hycgan w.v. I
‘think be mindful consider’ OFris. hugia w.v. I ‘think’ OS huggian w.v. I ‘id.’ OHG
huggen w.v. I ‘remember hope’. Further etymology uncertain. One of the most promising
attempts of a comparison of PG *hugi- and *hugjana- with extra-Germanic material is that
of Hirt (1900: 110), viz. that they are almost exact cognates to Skt. śúc - adj. ‘shining
bright pure’ and ś c át v. ‘radiates shines’ the only irregularity in the cognateness being
that Skt. śúc - (< PIE *k̂úk-i-) is the radically stressed adjective/agent noun and PG *hugi-
(< PIE *k̂ k-í-) is the suffixally stressed abstract noun formed from the former by means of
internal derivation. The semantic differences between the Indic and Germanic forms can
easily be overcome: The semantic field of intelligence, knowledge and mind often
combines well with notions of brightness and light, cf. e.g. the English use of brilliant and
bright. Literature: Bammesberger (1990: 130), Casaretto (2004: 188-189), Holthausen
(1974: 183), Kroonen (2013: 252), Lehmann (1986: 192-193), Orel (2003: 190-191),
Philippa et al. (2005: 428), Sehrt (1966: 275-276), De Vries (1962: 265, 274).
103 PG *hupi- m. ‘hip’. Represented in Goth. hups m. ‘id.’ OE hype m. ‘id.’ OHG huf f. ‘id.’.
To be reconstructed as PIE *kub-i- to the root PIE *ke b- ‘lie down recline’ cf. also e.g.
Lat. c b v. ‘lie down recline’ cubitus m. ‘elbow’ cubitum n. ‘id.’ MW kyscit v. ‘sleeps’
(< PIE *k b-sk̂e-) and maybe Gr. κύβος m. ‘hollow above the hip on cattle’ which should
then, of course, be separated from κύβος m. ‘dice’. Literature: Bammesberger (1990: 130),
Bjorvand & Lindeman (2000: 391), Casaretto (2004: 186), Holthausen (1974: 184), IEW
(2005: 589-590), Kluge/Seebold (2002: 425), Kroonen (2013: 257), Lehmann (1986: 196),
LIV (2005: 357-358), Orel (2003: 194), Philippa et al. (2005: 430).
104 PG *hwali- (or *hwala-?) m. ‘whale’. Represented in ON hvalr m. ‘id.’ OE hwæl m. ‘id.’
OHG wal, hwal m. ‘id.’. To be reconstructed as PIE *kwól-i- to the root PIE *kwel- ‘turn’
with the orignal meaning probably being ‘turner roller; turning rolling’ due to the rolling
movement made by the back of the whale on the surface of the water when the whale
surfaces to breathe. Consequently, this noun should probably rather be equated with the i-
stem adjectives, originally being adjectival agent nouns, treated below (section 2.4).
Alternatively to be compared to OPr. kalis m. ‘wels catfish’ and further possibly Av. kara-
n. ‘kind of fish’ Lat. squalus m. ‘kind of big sea fish’ < PIE *(s)kwal-o- ‘large fish’.

101
Literature: Bjorvand & Lindeman (2000: 410-411), Holthausen (1974: 179), IEW (2005:
639-640, 958), Kluge/Seebold (2002: 969), Kroonen (2013: 262), LIV (2001: 386-388),
Orel (2003: 197), Philippa et al. (2009: 591), De Vries (1962: 268-269).
105 PG *kw n - f. ‘wife’. Represented in Goth. qens f. ‘woman wife’ ON kvæn, kván f. ‘id.’
OE cw n f. ‘woman wife queen prostitute’ OS q n f. ‘woman’. The word for ‘woman’
is probably to be reconstructed as a proterodynamic h2-stem, i.e. PIE *gwén-h2-s (nom.sg.)
~ *gwn-éh2-s (gen.sg.), cf. the coexistence of OIr. bé(n) f. ‘woman’ and mná f. ‘id.’.
Lengthened grade forms are not limited to Germanic, cf. e.g. Skt. -j n - f. ‘wife’ (only in
exocentric compounds). Darms (1978: 74-76) considers but, in contrast to Harðarson
(1987: 130-133), ultimately rejects the possibility of PG *kw n - f. ‘wife’ to be a vrddhi
derivative from PG *kwen n- f. ‘woman’ > Goth. qino f. ‘woman’ ON kona f. ‘wife’
kvinna f. ‘woman’ OE cwene f. ‘id.’ OS quena f. ‘woman wife’ OHG quena f. ‘id.’. He
prefers to analyse it as continuing a form from the ablauting paradigm of the PIE word for
‘woman’. Such a form i.e. PIE *gw n- vel sim., could simply be phonologically regular
from PIE *gwenh2 (PIE *-VRH# > *- ̄ ), cf. Jasanoff (1989: 136-140). Schaffner (2001:
425-426) presents another option viz. that the word for ‘woman’ is actually an acrostatic
/e-ablauting noun, i.e. PIE *gw n 2- ~ *gwénh2-. Literature: Bammesberger (1990: 128,
177, 203), Bjorvand & Lindeman (2000: 473-475), Casaretto (2004: 186-187, 230),
Holthausen (1974: 64, 66), IEW (2005: 473-474), Kroonen (2013: 316-317), Lehmann
(1986: 275-277), NIL (2008: 177-185), Orel (2003: 228, 230), Sehrt (1966: 429, 431), De
Vries (1962: 325, 338-339).
106 PG *rugi- m. ‘rye’. Represented in ON rugr m. ‘id.’ OE ryge m. ‘id.’. Safe comparanda
can be found only in Balto-Slavic, cf. e.g. Lith. r g aĩ m. ‘rye’ (pl.) Latv. rudzi m. ‘id.’
(pl.), OCS rъžь f. ‘id.’. Kroonen (2013: 416) speculates if this post-PIE *r ĝhi- could be of
substrate origin, cf. also the semantically closely related Thrac. βρίζα f.(?) ‘emmer-wheat,
rye’ which cannot be compared to this post-PIE *r ĝhi- unless substrate influence is
invoked or unless it can be safely assumed that PIE * r gh- e 2-? > Thrac. βρίζα.
Literature: Bjorvand & Lindeman (2000: 738), Holthausen (1974: 265), IEW (2005: 1183),
Kluge/Seebold (2002: 769), Kroonen (2013: 416), Orel (2003: 308), De Vries (1962: 453).
107 PG *segi- m. ‘victory’. Represented in ON sigr ‘id.’ OE sige m. ‘id.’ OFris. si m. ‘id.’
OS sigi m. ‘id.’ OHG sigi m. ‘id.’. Radical full grade also in e.g. PG *segu- m. ‘victory’ >
OHG sigu m. ‘id.’ (with radical vocalism taken over from sigi) and PG *sigiz- n. ‘victory’

102
> Goth. sigis n. ‘id.’ (inflected as an a-stem), OE sigor m. ‘id.’. In reality all the
aforementioned Germanic forms may have originally belonged to the s-stem PG *sigiz-
with extra extra-Germanic cognates in Skt. sáhas- n. ‘force victory’ Av. hazah- n. ‘act of
violence robbery’ and further Skt. sáhate v. ‘defeats conquers’ Gr. ἔχω v. ‘have hold
possess’ MIr. seg m. ‘strength’; all ultimately derived from the root PIE *seĝh- ‘hold
possess overcome victory’. For the reanalysis of s-stems as i-stems cf. PG *sweli- m.
‘callosity callous skin’. Literature: Bammesberger (1990: 136), Bjorvand & Lindeman
(2000: 768-769), Boutkan & Siebinga (2005: 338), Casaretto (2004: 562), Holthausen
(1974: 293), IEW (2005: 888-889), Kluge/Seebold (2002: 847), Kroonen (2013: 430),
Lehmann (1986: 302), LIV (2001: 515-516), Orel (2003: 322), Philippa et al. (2009: 654),
Sehrt (1966: 463), De Vries (1962: 474).
108 PG *spurdi- m./f. ‘track course’. Represented in Goth. spaurds f. ‘id.’ OE spyrd m. ‘id.’
OHG spurt m. ‘id.’. In the light of its closest cognates Skt. s rd - f. ‘contest fight’ Av.
s әrәd- f. ‘zeal alacrity’ which are both root nouns (< PIE *s rdh-), it seems reasonable to
claim that PG *spurdi-, too, was originally a root noun, i.e. PG *spurd-. Such an
assumption is supported by the fact that Goth. spaurds is feminine, a feature seldom
observed in Germanic primary i-stems but richly present in root nouns, cf. also the
feminine gender of the Indo-Iranian cognates. The West Germanic forms would then have
changed their gender in order to conform to the general pattern of the i-stems. For further
analysis of this lexeme, cf. Ph.D. article no. 1. Literature: Casaretto (2004: 41-42),
Griepentrog (1995: 367-379), Holthausen (1974: 314), IEW (2005: 995-996), Kroonen
(2013: 470-471), Lehmann (1986: 319), LIV (2001: 580-581), Orel (2003: 367).
109 PG *þuli- m. ‘reciter’. Represented in ON þulr m. ‘speaker orator; sage’ OE þyle m.
‘orator spokesman’. Radical zero grade also in e.g. PG *þ l n- f. ‘recitation’ > ON þula f.
‘series of words’. Serves as the basis for the denominal verb PG *þuljana- w.v. I ‘speak
recite murmur’ > ON þylja w.v. I ‘id.’. Further etymology uncertain but probably either
to be compared to Hitt. talliyazi v. ‘prays to evokes’ and thus to be reconstructed as PIE
*tlH- in which case, however, the competing form ON þauli m. ‘reciter’ (< PG *þaulan-)
would need to be analysed as displaying secondary ablaut, or to be equated with Hitt.
tuliya- c. ‘gathering assembly’ for which Kloekhorst (2008: 897-898 with. lit.) suggests a
precursor PIE *tuh2-l- o- from the root PIE *te 2- ‘swell become strong’ cf. also Lith.
tūlas adj. ‘many’ < PIE *tuh2-lo-. The short vowel of PG *þuli- and its derivatives (< PIE

103
*tuh2-li-) would thus have to be explained by means of Dybo’s Law of pretonic shortening.
In any case, given the validity of either analysis presented here, PG *þuli- cannot be
included in the list of primary Germanic i-stems. Only if the frequently mentioned, yet
semantically weak comparison of PG *þuli- ‘reciter’ and PG *þ l ja na- w.v. III ‘tolerate
endure’ (< PIE *tl 2-éh1-) is invoked, will any analysis of PG *þuli- be possible.
Literature: Bammesberger (1990: 133), Holthausen (1974: 374), Kroonen (2013: 535,
550), Orel (2003: 428), De Vries (1962: 626, 630).
110 PG *w d - f. ‘clothes, clothing’. Represented in ON váð f. ‘woven fabric cloth piece of
stuff’ OE w d, w de f. ‘article of dress’ OFris. w d f. ‘garment’ OS w d, w d f./n.
‘id.’ OHG w t f. ‘garment; armour’. Further etymology uncertain but the possibility
exists for PG *w di- to be compared to Lith. áusti v. ‘weave’ (pres. á dž a ), ūd s f.
‘woven fabric’ and thus be reconstructed as PIE *h2 e 1dh-i- to the root PIE *h2 e 1dh-
‘weave’ of which the Lithuanian form would continue a Schwebe-ablauted variant PIE
h
*h2e 1d -. The idea that PG *w d - should be connected to the class V strong verb PG
*wedana- s.v. ‘conjoin bind’ is rejected by i.a. Lehmann (1986: 154). Literature:
Bammesberger (1990: 136), Bjorvand & Lindeman (2000: 1008-1009), Boutkan &
Siebinga (2005: 433), Holthausen (1974: 378), IEW (2005: 75-76), Orel (2003: 460),
Philippa et al. (2005: 259-260), Philippa et al. (2009: 586), Sehrt (1966: 633), De Vries
(1962: 637).
111 PG *wr g - f. ‘complaint accusation’. Represented in Goth. wrohs f. ‘id.’. Radical
lengthened o-grade also in e.g. PG *wr ga- n. ‘quarrel’ > ON róg n. ‘quarrel calumny’.
Further etymology uncertain. Literature: Bammesberger (1990: 136), Casaretto (2004:
189-190), Kluge/Seebold (2002: 774-775), Lehmann (1986: 411), Orel (2003: 473), De
Vries (1962: 450).
112 PG *wurmi- f. ‘worm’. Represented in OE wyrm m. ‘snake worm dragon’ OFris. wirm
m. ‘worm’ OS wurm m. ‘id.’ OHG wurm m. ‘worm snake’. Radical zero grade also in
e.g. PG *wurma- m. ‘worm’ > Goth. waurms m. ‘snake’ ON ormr m. ‘id.’. Extra-
Germanic cognates abound, cf. e.g. Gr. ῥόμος m. ‘wood borer woodworm’ Lat. vermis m.
‘snake’ Lith. var as m. ‘insect mosquito’ ORu. vermije n. ‘insects’ (pl.); all from PIE
* r -, * r o- or secondarily * or o- with a rhyming parallel in PIE *kwr - ‘worm
grub’ (> Skt. k - m. ‘worm insect’ Lith. kirmìs m./f. ‘grub’ etc.) which is in turn
probably affiliated with the root PIE *kwremH- ‘take a step’ reflected in e.g. Skt. kr at v.

104
‘strides paces’. Literature: Bjorvand & Lindeman (2000: 691-692), Casaretto (2004: 383),
Holthausen (1974: 412), IEW (2005: 649, 1152), Kluge/Seebold (2002: 998), Kroonen
(2013: 600), Lehmann (1986: 397), LIV (2001: 368-369), Orel (2003: 476), Philippa et al.
(2009: 637-638), Sehrt (1966: 725), De Vries (1962: 420).

Not mentioned in the previous paragraphs are the Germanic i-stem ethnonyms of the type Goth.
saureis m. ‘Syrians’ (pl.) ON danir m. ‘Danes’ (pl.) firðir m. ‘people of Fjǫrð’ (pl.) OE dene m.
‘Danes’ (pl.) engle m. ‘Angles’ (pl.) cf. e.g. Krahe & Meid (1967: 67). Although isolated in the
overall derivational and inflectional system of Germanic, this type finds a parallel outside
Germanic, viz. above all in the Indo-Iranian i-stem patronyms, and must thus be expected to be of
Proto-Indo-European age cf. e.g. the vrddhi’ed examples of Skt. gn veś - (to agn veśa-) and
śa c v k - (to ś c v k a-) mentioned by Wackernagel & Debrunner (1954: 301-302).

2.4. I-stem adjectives

Of mere marginal relevance to the present study, which focuses on the formation of nouns only, are
the Germanic primary i-stem adjectives, cf. e.g. Krahe & Meid (1967: 66-67) and Bammesberger
(1990: 259-261). I-stem adjectives fall into two functional groups in Germanic, viz. gerundives, i.e.
adjectives of possibility, on the one hand as exemplified by PG *flugi- adj. ‘able to fly’ (derived
from PG *fleugana- s.v. ‘fly’) and PG * t - adj. ‘eatable’ (derived from PG *etana- s.v. ‘eat’) and
adjectival agent nouns on the other as illustrated by PG *swiki- adj. ‘deceiving’ (derived from PG
*swīkana- s.v. ‘deceive’) and PG *t - adj. ‘suitable proper’ (derived from PG *temana- s.v. ‘be
proper’). Formally speaking, both functional types tend to appear with radical zero grade when
derived from class I-III strong verbs and with radical lengthened grade as found in the stem of the
preterite plural when derived from class IV-VI strong verbs.25
Counterexamples appear to occur, though, but closer scrutiny reveals that all but one are, in fact,
regular. PG *muni- adj. ‘remembering’ derived as it is from a class IV strong verb should regularly
appear as PG † n - if it were to follow the pattern of other i-stem adjectives derived from class IV

25 In my view, it is quite feasible that the lengthened grade type has originally arisen as a result of vrddhi. After all,
with their general function of designating ‘pertaining to X relating to X’ adjectives constitute the ideal environment
for vrddhi to occur. Just as the i-stem nouns, these i-stem adjectives could easily be adapted to the ablaut system of
the Germanic strong verbs, but whereas the nouns are closely connected to the stem of the preterite participle, the
adjectives must be analysed as corresponding to the ablaut grade of the stem of the preterite plural.

105
strong verbs. It is important to bear in mind, though, that the verb in question is no ordinary strong
verb but a preterite-present one, viz. PG *man ~ *munum pp.v. ‘think’ whose ablaut pattern
deviates slightly from that of the regular strong verbs.
The motivation for PG *baugi- adj. ‘flexible pliant’ (derived from PG *beugana- s.v. ‘bend’)
not to appear as PG †bugi- might be identical to the explanation provided for PG *laudi- m. ‘form
appearance’ viz. that it is only attested as a second member of an exocentric compound in which
case transition of thematic stems (a-stems) into i-stems is, from a Proto-Indo-European point of
view, an expected and regular process.
Only PG *kausi- adj. ‘choosy’ (derived from PG *keusana- s.v. ‘test trial select prefer’)
completely resists analysis within the framework of the posited structure of the adjectival i-stems by
displaying both synchronically aberrant radical o-grade and unexpected unvoiced Verner’s variant
of the root-final consonant. First speculations inevitably focus on the possibility of PG *kausi-
being of the same nature as PG *baugi- adj. ‘flexible pliant’ and PG *laudi- m. ‘form appearance’
viz. a second member of an exocentric compound, but contrarily to these, PG *kausi- adj. ‘choosy’
represented by OE cīes, cīese adj. ‘id.’ only appears as simplex whereas the related a-stem adjective
PG *kausa- represented by OE (or- c as adj. ‘invulnerable’ and (un-be- c as adj. ‘indisputable
incontrovertible’ only exists as a second member of compounds. Consequently the only analysis
that presents itself so far is to regard PG *kausi- as an archaism.

3. Form and function of Indo-European primary i-stems

The highly heterogenous and partly inexplicable appearance of Germanic i-stems uncovered in the
previous chapter emphasises the need for inclusion of data from the other Indo-European branches.
Only then will we find ourselves capable of determining which forms and functions are archaic,
which are productive, and which are neither.
Consequently, we shall now turn our attention towards those of the Indo-European branches
where primary i-stems have entirely or partially remained a separate category, cf. also Hinderling
(1967: 104-112).

3.1. Indo-Iranian (represented here by Sanskrit)

As ingeniously analysed by Wackernagel & Debrunner (1954: 291-307) who is followed by e.g.
Hinderling (1967: 104-106), we may categorise the primary i-stems of Sanskrit into two main
groups containing two subgroups each, viz. agent nouns and action nouns.

106
The group of agent nouns consists of verbal adjectives with the root in the zero grade preceded
by a stressed reduplicative syllable, cf. e.g. Skt. cákri- adj. ‘working’ and true agent nouns with a
stressed radical vowel mainly in the zero grade. Examples of this latter type include Skt. dhúni- adj.
‘roaring sounding’ (derived from dhvan-) and - m. ‘singer’ (derived from - ‘go’). What appear
to be radical full or o-grade i-stems are attested, too, cf. e.g. Skt. go-dari- adj. ‘opening the stables
of the sky’ and vasu-váni- adj. ‘asking wealth bestowing wealth’ but in such cases the final Skt. -i
reflects PIE *-H of se -roots with the consequence of the nouns in question resisting analysis as i-
stems and being, in fact, mere root nouns, cf. also Skt. vasu-ván- adj. ‘asking wealth bestowing
wealth’ competing with vasu-váni- adj. ‘id.’.
Whereas the former i-stem agent noun subtype is unknown to any other branch of Indo-
European, strong parallels can be drawn from the latter type to the Germanic i-stem adjectival agent
nouns mentioned in section 2.4 the only caveat being that Verner’s Law points at all Germanic i-
stem adjectival agent nouns safe for PG *kausi- adj. ‘choosy’ continuing Proto-Indo-Europan
suffixal accent.
Constituting the second major group of Sanskrit primary i-stems, action nouns, too, are divisible
into two subgroups, viz. fossilised dat.sg. forms created from the unstressed zero grade of the root
and functioning synchronically as infinitives on the one hand, cf. e.g. Skt. d śá e v. ‘see’ and citáye
v. ‘understand’ and true agent nouns with more or less unpredictable gender, accent and radical
ablaut grade on the other, cf. e.g. Skt. k í- f. ‘agriculture farming; ploughing’ bhují- f. ‘granting of
enjoyment favour’ śocí- f. ‘flame glow’ añjí- m. ‘ointment’ (also as adj. ‘applying an ointment or
pigment’) and rúci- f. ‘light lustre splendour beauty’. A strong tendency is seen though for this
latter subtype to display radical zero grade and, when the action noun keeps some of its verbal
semantics, feminine gender. Common reason dictates that this type has a strong connection with the
Germanic i-stem verbal abstracts, differing from those mainly by the general application of
feminine gender in Sanskrit versus masculine gender in Germanic.
Sanskrit and Germanic thus seem to have two common points of reference as regards the primary
i-stems, viz. agent nouns in Sanskrit (stressed on the root syllable appearing in the zero grade)
compared to adjectival agent nouns in Germanic (mainly stressed on the i-stem suffix and mainly
with radical zero grade) and infinitives and action nouns in Sanskrit (mainly stressed on the i-stem
suffix and with unpredictable radical ablaut grade) compared to action nouns, i.e. verbal abstracts,
in Germanic (stressed on the i-stem suffix and with unpredictable radical ablaut grade).

107
As a final point of relevance, it should be noted that Sanskrit and Germanic match each other in
one more area regarding the i-stems. Sanskrit gerundives can be formed by the addition of the suffix
Skt. -yá- to a verbal root most often in the zero grade, cf. Wacernagel & Debrunner (1954: 789-
795), thereby bearing a strong resemblance to the i-stem gerundives in Germanic of the types PG
*flugi- adj. ‘able to fly’ and PG * t - adj. ‘eatable’ cf. section 2.4. Furthermore the fact that the
Germanic i-stem adjectives of both functional types are also frequently formed with radical
lengthened grade finds a perfect parallel in Sanskrit, cf. e.g. the comparison of Skt. s d - m. ‘rider
horseman’ (originally in compounds only) and Skt. s d á- adj. ‘fit for riding’ m. ‘riding horse’
with PG *s t - adj. ‘sitting; able to sit’ i.e. both adjectival agent noun/adjective and gerundive cf.
e.g. Wacernagel & Debrunner (1954: 295) and Heidermanns (1993: 479-480).26

3.2. Greek

Only few parallels can be drawn to the i-stems of Germanic and Sanskrit if we choose to accept
Solmsen’s (1909: 155-179) categorisation of Greek primary i-stems into abstracts with radical
stressed zero grade on the one hand, cf. e.g. Gr. σπάνις f. ‘scarcity lack’ (maybe belonging to Gr.
πένομαι v. ‘toil work; am poor am needy’ cf. e.g. Frisk (1963-1966: 756-757)) as well as a minor
group of i-stems with even less certain etymologies, and concrete nouns with radical stressed o-
grade on the other, cf. e.g. Gr. στρόφις f. ‘slippery fellow twister’ τρόπις f. ‘keel’ and τρόχις m.
‘courier messenger’ on the other.
According to Chantraine (1933: 111), all types of Greek i-stems have remodelled the original
accent into a new system. Hence follows that the location of the accent in a given Greek i-stem
reveals virtually nothing about the conditioning of Proto-Indo-European i-stem accentuation. The
question now remains to be answered if the categorisation suggested by Solmsen (1909: 155-179)
can be upheld or if amendments are needed. Solmsen himself (1909: 162) actually doubts that his
proposed categorisation mirrors completely any older stages. Firstly, he points out, other o-grade
derivatives may have influenced on the i-stem concrete nouns with radical o-grade. Secondly, the
very assignment of zero grade i-stems in the group of abstract nouns and of o-grade i-stems in the
group of concrete nouns is far from consistent, cf. e.g. the i-stem abstract noun Gr. φρόνις f.
‘prudence wisdom’ with radical o-grade which would be unexpected according to Solmsen’s
categorisation. To this latter objection might be added that, given the appurtenance of Gr. πόρις f.

26 It cannot be rouled out that at least Skt. s d - and PG *s t - are not as similar as we are first led to think. Besides
being reconstructable as PIE *s d -, Skt. s d - may, unlike PG *s t -, alternatively continue PIE *sódi-.

108
‘calf young heifer’ to the root PIE *perh3- ‘bear produce’ cf. e.g. IEW (2005: 818) and LIV
(2001: 474) it is actually to be understood as an abstract or action noun i.e. ‘the born one the
begotten one’ rather as a concrete or agent noun i.e. ‘bearer begetter’. Also it cannot be ruled out
that Gr. ἄγυρις f. ‘gathering crowd’ may be a mere Aeolic i-stem variant of the more wide-spread
form ἀγορά f. ‘assembly’ cf. the development in Aeolic of Gr. -ορ- > -υρ- mentioned by Schwyzer
(1959: 351-352), in which case yet another abstract noun with underlyingly radical o-grade may be
attested.
Even if Solmsen (1909: 155-179) is right as to the general tendencies in the synchronic i-stem
system of Greek, it lies beyond dispute that a different system must be reconstructed for the
precursors of Greek, viz. a system with seemingly random assignment of radical o- and zero grade
in both functional i-stem types and, consequently, with stronger affinities to the systems of
Germanic and Sanskrit. From a diachronic point of view, an even clearer pattern emerges if the
nouns in Gr. -ἰς, -ί - of the type Gr. ῥανίς f. ‘drop spot’ are drawn into the equation, as well, cf. e.g.
Chantraine (1933: 338) and Schwyzer (1959: 464). As suggested by the frequent coexistence of id-
and i-stem forms, we may expect at least some of these nouns to be old, primary i-stems that were
reanalysed as id-stems only at a late stage in the history of the Greek language. These original i-
stem nouns with radical zero grade and suffixal accent are probably to be compared primarily to the
i-stem infinitives and action nouns of Sanskrit as well as to the i-stem verbal abstracts of Germanic.
Probably secondarily, however, they are also to be compared to the (adjectival) agent nouns of
Sanskrit and Germanic seeing that agentive function is also attested, cf. e.g. Gr. σκαφίς f. ‘spade
showel; belly bunt’ i.e. ‘digger digging tool’ to σκάπτω v. ‘dig’ and Gr. παγίς f. ‘trap snare’ to the
root PIE *peh2k̂- ‘fix fasten’.
The i-stems of both functional types with radical o-grade deserve an additional comment. If Gr.
ἄγυρις is not to be analysed as an o-grade formation but, as often most done, as a zero grade
formation, we cannot help noticing that, in all the o-grade forms attested, the radical vowel is
preceded by an r. Consequently, the theoretical possibility exists that these i-stems all display
Aeolic zero grade, i.e. -ρο- for standard Gr. -ρα- < PIE *-r-. The reason why I do not consider that
an option, after all, is that we would expect at least one of these forms at least once to occur with
the standard vocalism Gr. -ρα-. Seeing that this is not the case, I, albeit aware that radical zero grade
is an option, prefer the analysis of these Greek i-stems as o-grade formations.

109
3.3. The remaining branches

In my view, Indo-Iranian and Greek provide us with the best comparative evidence for the
Germanic primary i-stems. However, we should not neglect that the remaining branches have i-
stems as well.
Hittite, as a representative of the Anatolian branch, offers two formal types of i-stems, viz. one
type with suffixal -i- and one type with suffixal -ai- in the strong stem of the nom.sg. and acc.sg.,
and, as is the case in Germanic, both nouns and adjectives are attested, cf. e.g. Kronasser (1962-
1966: 202-208).
In Tocharian, i-stems are recognisable by their nom.pl. -i which has a palatalising effect on the
preceding consonant; the i-stem suffix itself is apocopated. According to Van Windekens (1979: 11-
12) and indirectly Adams (1988: 125-126), Tocharian has preserved only a small residue of the
primary i-stems. Furthermore, we may easily risk mistaking an i-stem for an ī-stem seeing that their
nom.sg. would both disappear by means of apocopy. In the acc.sg., however, we might still have a
chance of keeping them apart, cf. again van Windekens (1979: 12).
Armenian contains only a couple of archaic i-stems, i.e. Armenian i-stems that correspond to i-
stems in other Indo-European languages, cf. Olsen (1999: 78).
In Latin, the i-stem inflection and the consonant stem inflection have merged into one paradigm,
for which reason any attempt of deducing evidence from the Latin i-stems that might be of value to
the comparative study of Proto-Indo-Euopean i-stems is severely hampered, cf. e.g. Leumann et al.
(1963: 231-233). In contrast to Latin, the Sabellic languages manage to keep consonant stems and i-
stems inflectionally apart, cf. Buck (1904: 124-127), but due to their relatively small number, they
do not contribute with much in comparative terms.
By consulting some of the standard handbooks on Balto-Slavic nominal word formation, e.g.
Otrębski (1965: 49-53) and Vaillant (1974: 22-23), we learn that, in Baltic as well as in Slavic, the
i-stems have considerably increased in number due to the transition of, above all, root nouns to the
inflectional class of the i-stems, cf. also Larsson (2010: 34-35, 49-100, 102). In both branches,
however, zero grade i-stems still seem to play a certain role, cf. Hinderling (1967: 110-112) and e.g.
the comparison of OCS lъžь adj. ‘lying false mendacious’ and PG *lugi- m. ‘lie deception’ (even
if the former is an adjective and the latter is a noun).

3.4. Proto-Indo-European

As can be deduced from sections 3.1-3.3, Anatolian, Tocharian, Armenian, Italic and Balto-Slavic
appear to offer i-stem forms of less informative value to the purpose of this study than do Indo-

110
Iranian and Greek. Consequently, the following brief outline of the Proto-Indo-European state of
affairs is based mainly on Indo-Iranian, Greek, and Germanic.

3.4.1. General types of i-stems in Proto-Indo-European

When comparing the i-stem types found in the branches mentioned above, we find two prevalent
types, viz. adjectival agent nouns and action nouns often functioning as verbal abstracts.
Adjectival agent nouns have a clear tendency in Sanskrit to be stressed on the root; so, too, in
Greek where, however, radical accent also prevails in the abstract noun type. Where a difference
can be seen, Germanic points at suffixal accent in all i-stem forms but one, viz. PG *kausi- adj.
‘choosy’. As for the radical ablaut grade Indo-Iranian is less informative due to its coalescence of
the primary vowels PIE *e, *a and *o > PIIr. *a.27 It is beyond dispute, however, that radical zero
grade is widely spread in both main i-stem types. Germanic and Greek reveal what Indo-Iranian
cannot do or can do only to a limited extent, viz. that the other main ablaut grade found in this type
of i-stems is the o-grade.
As for the action nouns or verbal abstracts, there is a strong tendency for them to be stressed on
the i-stem suffix, albeit with the sole exception of a few forms in Greek and Sanskrit. Again, all
three branches reveal that both radical o- and zero grade is attested even if zero grade seems to be
the by far most frequent type, cf. especially the widely productive infinitives in Skt. CC-áye. When
it comes to the gender of these i-stems, however, any agreement between the branches in question
ceases completely: Greek and Sanskrit show a clear preference for feminine gender, more
conspicuously so in Greek than in Sanskrit, whereas the bulk of Germanic i-stem verbal abstracts
comes with masculine gender.

3.4.2. Derivational history of primary i-stems with radical zero grade

Besides being acknowledged as a Proto-Indo-European derivational type due to the cognateness of

the Indo-Iranian, Greek and Germanic primary i-stems with radical zero grade, this type remains
isolated within Proto-Indo-European in that it, at least at first glance, resists adaption into the
general derivational system.

27 In two positions the timbre of the PIE vowel actually does reveal itself in Indo-Iranian viz. where Brugmann’s Law
causes lengthening of PIE *o > PIIr. * in open syllables, cf. e.g. Wackernagel & Debrunner (1896: 13-14), and
where a PIE *e (not *a or *o) palatalises a preceding velar plosive, cf. e.g. Wackernagel & Debrunner (1896: 139-
144).

111
The possibilty exists, however, for these zero grade i-stems as well as for i-stems with other
radical ablaut grades to be derived or reshaped from old root nouns, cf. e.g. Hinderling (1967: 113-
115). To what extent that process can be regarded as having taken place in the Indo-European
proto-language remains uncertain, but if we assume that this process of transferring old root nouns
into i-stems has its origin in compounds with a root noun as their second member, a thematisation
or adjectivisation of the root noun would, if early enough, be expected to appear as PIE *-i-, cf. e.g.
Brugmann & Delbrück (1906: 112-113) and Rasmussen (1988 [1999]: 320). On the other hand,
phonological motivations for a transition from root nouns to i-stems can be identified in (some of)
the individual branches cf. e.g. the Sanskrit root nouns from se -roots ending in -i (Wackernagel &
Debrunner 1954: 294-295) or the Latin and Balto-Slavic coalescence of the acc.sg. of root nouns
(PIE *- > Lat. -em, -im, PBS *-im) with the corresponding desinence of the i-stems (PIE *-im >
Lat. -im, PBS *-im) (Sihler 1995: 316-318; Larsson 2010: 34-35; Vaillant 1958: 146, 182-183), thus
suggesting the transitions to be of a more recent date. Of course, these two alternatives by no means
exclude each other. To put it differently, the seeds of the transitions may have been sown in Proto-
Indo-European already, but the phonological motivations in some of the daughter languages have
promoted the process of changing root nouns into i-stems even further.

3.4.3. Derivational history of primary i-stems with radical ο-grade

In the light of Rasmussen’s (1988 [1999]: 313) claim that the thematic vowel PIE *e ~ *o was
originally weakened into PIE *i ~ *u when unstressed, speculations as to whether the primary i-
stems with radical o-grade could not simply be variants of the widespread o-grade thematic nouns
and adjectives of the type τόμος/τομός seem not too far-fetched. Consequently, Rasmussen (1988
[1999]: 320) himself was not slow to suggest that option. Also Casaretto (2004: 173) suggests a
connection, albeit independent of any regular sound change of PIE *e ~ *o > *i ~ *u, between
Germanic i-stems of the PG *balgi- type and the PIE τόμος type.
However, if the primary i-stems are to be included into the derivational conglomerate of
τόμος/τομός and also τομή, they need to fit into the derivational chain of these possibly related
formations, for which cf. e.g. Risch (1974: 196-207) and Rasmussen (1989: 156-158). If taking
feminine abstract nouns of the type Gr. τομή f. ‘cutting incision insection’ i.e. PIE *CoC-éh2-, as
our point of departure, we can, by using that form as a second member of exocentric compounds,
obtain the type Gr. X-τομός, -ή, -όν adj. ‘relating to cutting of X’ which after becoming separated
from its original compound, gives rise to agentival adjectives with suffixal stress of the type Gr.
τομός, -ή, -όν adj. ‘cutting sharp’ i.e. PIE *CoC-ó-. With internal derivation by means of

112
contrastive accent, the creation of an action noun of the type Gr. τόμος m. ‘slice cut’ i.e. PIE
*CóC-o-, from Gr. τομός, -ή, -όν is a completely straightforward process.28
Further, we may tentatively hypothesise, following Rasmussen (1988 [1999]: 320; 1989: 156),
that the process of internal derivation yielding Gr. τόμος above was preceded by a much earlier
process of internal derivation at a time when Rasmussen’s rule stating that the thematic vowel PIE
*e ~ *o was weakened into PIE *i ~ *u in unstressed position was still active. By that process, we
may obtain a shift of word class from the agentival adjectives with suffixal stress of the type Gr.
τομός, -ή, -όν to true agent nouns of the type PIE *CóC-o- > *CóC-i- with presumably masculine
gender. Finally, yet another process of internal derivation of PIE *CóC-i-  *CoC-í- would then
turn these agent nouns into action nouns or general verbal abstracts.

4. Primary i-stems as a parallel type of the PIE toga/fuga formations

Rasmussen’s (1989: 158-175) further idea that nouns of the type Gr. τομή or Lat. toga f. ‘toga’ are
actually identical to nouns of the type Lat. fuga f. ‘flight escape’ i.e. feminine abstract nouns of the
structure PIE *CC-éh2-, might be what is needed in order for the two formal i-stem types to be
united and thus for the type with radical zero grade to lose its status as a derivationally isolated type
and be adapted into the general derivational pattern of Proto-Indo-European.
More precisely Rasmussen suggests that nouns formed from roots of the structures PIE ToT,
RoT, HRoT, TRoT, sToUT, HRoUT, sRoUT, UoT, ToRH, ToR, sToR, HToR, TRoR, ToU, TRoU,
sRoU, TOUh1, sToUh1, Tos, HUoRs, ToRT, sToRT, RoRT, TRoRT, sTRoRT, sRoRT, HRoRT,
URoRT, soRT and UoRT keep the PIE *o of their root, whereas nouns formed from roots of the
structures PIE T-UT, R-UT, H-UT, TR-UT, sR-T, H-RH, U-RH, T-Uh2/3, sT-Uh2/3, U-UH, TR-TH?,
T-ST, H-U, TT-H, TU-H, HR-H, C-HU, C-RHU and sU-RT lose it.
Seeing that the i-stem nouns with radical o-grade may actually concord well with the general
τόμος/τομός/τομή system, we would a priori expect them to concord equally well with the
complementary distribution of o- and zero grade prevailing in at least the toga/fuga type, the

28 Widmer (2004: 32-33), however, believes the derivational process of the τόμος/τομός type to have run in the reverse
order, i.e. Gr. τόμος m. ‘slice cut’  τομός, -ή, -όν adj. ‘cutting sharp’. He basically analyses the latter as a
possessive derivative of the former by means of not true internal derivation but rather suffixal substitution of the
type ROOT-o1-  ROOT-ó2[+poss]- with a special possessive suffix PIE *-ó2[+poss]- also used for internal derivation
from athematic stems.

113
τόμος/τομός type being already in Proto-Indo-European times a highly productive type subject to
levelling in favour of keeping the PIE *o regardless of radical phonotactics, cf. Rasmussen (1989:
157-158). In tables 2-6 below, I list all the Germanic i-stem nouns analysed in this article in
accordance with their level of adaption to the toga/fuga type.

4.1. I-stems of the toga/fuga type

Table 2: Radical zero grade expected in the toga/fuga system

Derivational match with the ablaut grade of the stem of No derivational match with the ablaut grade of the stem
the preterite participle of a corresponding strong verb of the preterite participle of a corresponding strong verb

(1) PG *biti- m. ‘bite prick’ (102) PG *hugi- m. ‘intellect mind’

(2) PG *bliki- m. ‘appearance emergence’ (103) PG *hupi- m. ‘hip’

(3) PG *gridi- f. ‘step standing’ (106) PG *rugi- m. ‘rye’

(6) PG *lidi- m. ‘going’

(21) PG *flugi- m. ‘escape’

(23) PG *guti- m. ‘filling pouring’

(24) PG *hluti- m. ‘share lot’

(27) PG *lugi- m. ‘lie deception’

(29) PG *ruki- m. ‘smell; smoke’

(34) PG *tugi- m. ‘pull draw’

(47) PG *swulgi- m. ‘drink swallow gulp’

(49) PG *swulti- m. ‘starvation’

(50) PG *wunni- m. ‘suffering pain’

(69) PG *fal(l)i- m. ‘fall crash’

If our working hypothesis that the Germanic i-stem verbal abstracts mainly displaying radical zero
and o-grade ablaut and synchronically matching the ablaut grade of the stem of the preterite
participle of a corresponding strong verb actually have developed from only one archaic type of
suffixally stressed i-stem action nouns with complementary distribution of o- and zero grade
according to radical phonotactics is valid, the numerous forms listed in the left column of table 2 are
clearly the ones serving as pivots for the reinterpretation.

114
Needless to say since not all of the Germanic “roots” represented in these i-stems continue pure
Indo-European roots, cf. e.g. the proposed etymologies of (24) PG *hluti- m. ‘share lot’ (47) PG
*swulgi- m. ‘drink swallow gulp’ and (69) PG *fal(l)i- m. ‘fall crash’ the productively created
forms must have arisen in analogy with potentially inherited ones such as (1) PG *biti- m. ‘bite
prick’ (2) PG *bliki- m. ‘appearance emergence’ etc.
Isolated and unaffected as they are by the ablaut system of any strong verbs, the three forms of
the right column of table 2, except for (106) PG *rugi- m. ‘rye’ for which substrate influence may
have played a role, must be regarded as continuing truly archaic i-stems of the type PIE *CC-í-.

Table 3: Radical zero grade unexpected in the toga/fuga system

Derivational match with the ablaut grade of the stem of No derivational match with the ablaut grade of the stem
the preterite participle of a corresponding strong verb of the preterite participle of a corresponding strong verb

(4) PG *hrini- m. ‘touch’ (108) PG *spurdi- m./f. ‘track course’

(7) PG *siki- m. ‘sigh’ (109) PG *þuli- m. ‘reciter’

(9) PG *slidi- m. ‘misstep slip; error’

(10) PG *sliki- m. ‘furrow’

(11) PG *sliti- m. ‘break split’

(12) PG *snidi- m. ‘cut’

(13) PG *stigi- m. ‘ascent rising’

(14) PG *stiki- m. ‘prick stab’

(30) PG *skuti- m. ‘shot shooting’

(31) PG *slupi- m. ‘slip; piece of garment to slip over

one’s head’

(35) PG *brungi- m. ‘bringing’

(38) PG *drunki- m. ‘drink’

(39) PG *dunti- m. ‘shot’

(40) PG *hwurbi- m. ‘way passage’

(41) PG *kurbi- m. ‘fragment bit piece’

(43) PG *sprungi- m. ‘jump’

(44) PG *stungi- m. ‘prick stab’

115
(45) PG *stunkwi- m. ‘smell’

(46) PG *sturki- m. ‘strength’

(51) PG *wurdi- f. ‘fate destiny chance’

(52) PG *bruki- m. ‘breach crack fragment’

(53) PG *buri- m. ‘son progeny’

(54) PG *drupi- m. ‘blow stroke’

(55) PG *kuli- m. ‘cold’

(56) PG *kwumi- m. ‘coming’

(59) PG *stuni- m. ‘groan’

As can be deduced from table 3, a wide array of i-stem verbal abstracts appear with a non-expected
radical zero grade, i.e. a radical zero grade in roots of a structure where PIE *o would have been
expected instead.
The many forms of the left column of table 3 are in no need of further explanations, for they may
easily be analysed as productively derived with the ablaut grade found in the stem of the preterite
participle of a corresponding strong verb. Their very etymologies reveal that many of them are,
indeed, secondary. Suffice it here to mention e.g. the etymologies of (4) PG *hrini- m. ‘touch’ (38)
PG *drunki- m. ‘drink’ (43) PG *sprungi- m. ‘jump’ and (44) PG *stungi- m. ‘prick stab’ in the
roots of which nasal infixes have been inserted with the consequence that the i-stems do not
continue true Proto-Indo-European primary i-stems.
The two forms of the right column constitute potentially much greater problems to our working
hypothesis, seeing that there are no strong verbs in analogy with whose preterite participles such
zero grade forms, unexptected in the light of the rules governing the complementary distribution of
o- and zero grade in the toga/fuga system, could have been secondarily created or reshaped. As for
(108) PG *spurdi- m./f. ‘track course’ however we have already established that it was originally
a root noun rather than an i-stem in which case the rules governing the toga/fuga system cannot be
expected to apply, and the etymology of (109) PG *þuli- m. ‘reciter’ is simply too obscure in order
for it to constitute any strong counterexample.

116
Table 4: Radical o-grade expected in the toga/fuga system

Derivational match with the ablaut grade of the stem of No derivational match with the ablaut grade of the stem
the preterite participle of a corresponding strong verb of the preterite participle of a corresponding strong verb

(64) PG *kali- m. ‘cold’ (80) PG *smauki- m. ‘smoke’

(72) PG *stauti- m. ‘thrust push blow’ (81) PG *balgi- m- ‘sack bag’

(82) PG *bandi- m. ‘captivity’

(83) PG *dranki- m. ‘drink’

(84) PG *stangi- m. ‘bar pole staff stake’

(85) PG *stankwi- m. ‘smell’

(92) PG *drapi- m. ‘blow stroke’

(94) PG *mati- m. ‘food’

(99) PG *bak(w)i- m. ‘creek brook rivolet’

(101) PG *fadi- m. ‘lord’

(104) PG *hwali- m. ‘whale’

Parallel to the forms listed in the left column of table 2, the two forms of the left column of table 4
may theoretically have served as pivots for the reinterpretation of the archaic complementary
distribution of o- and zero grade according to radical phonotactics in the i-stem action nouns into a
new system where the radical ablaut grade of an i-stem action noun synchronically matches that of
the stem of the preterite participle of a corresponding strong verb. Only (72) PG *stauti- m. ‘thrust
push blow’ may prove original, though, seeing that (64) PG *kali- m. ‘cold’ has clearly been
formed secondarily to the, in itself, newly shaped preterite participle PG *kalena- ~ *kalana-
pret.ptc. ‘cold’ the archaic participial form being that of PG *kulena- ~ *kulana- pret.ptc. ‘feeling
shivery’ (< PIE *glH-enó- ~ *glH-onó-) represented in OSw. kolin adj./ptc. ‘id.’.
Truly archaic and relatively isolated examples of i-stems of the type PIE *CoC-í- are listed in the
right column of table 4. Note, however, that (104) PG *hwali- m. ‘whale’ is not to be compared to
the i-stem abstract noun type of PIE *CoC-í- but rather to the adjectival agent noun type of PIE
*CóC-i- if the whale is indeed to be regarded as ‘turner roller’ thereby matching the root PIE
*kwelh1- ‘turn roll’ from the o-grade of which (104) PG *hwali- is derived. Also, as is the case for
their synchronically regular zero grade counterparts (38) PG *drunki- m. ‘drink’ and (44) PG
*stungi- m. ‘prick stab’ listed in table 3 the roots of (83) PG *dranki- m. ‘drink’ and (84) PG

117
*stangi- m. ‘bar pole staff stake’ have been enlarged with a nasal infix which implies that these two i-
stems can in no way be regarded as archaisms, their conformity to the complementary distribution of o-
and zero grade in the toga/fuga system thus being merely coincidental.

Table 5: Radical o-grade unexpected in the toga/fuga system

Derivational match with the ablaut grade of the stem of No derivational match with the ablaut grade of the stem
the preterite participle of a corresponding strong verb of the preterite participle of a corresponding strong verb

(75) PG *hlauti- m. ‘share lot’

(76) PG *laudi- m. ‘form appearance’

(78) PG *rauki- m. ‘smoke’

(86) PG *swalgi- m. ‘abyss; swirl whirlpool’

(87) PG *swangwi- m. ‘swing; stroke’

(88) PG *swanki- m. ‘misery sorrow; toil; temptation’

The six forms listed in the right column of table 5 are, together with those two of the right column of
table 3, the ones that may justly contradict the working hypothesis presented in this article.
Three of the forms, however, viz. (86) PG *swalgi- m. ‘abyss; swirl whirlpool’ (87) PG *swangwi-
m. ‘swing; stroke’ and (88) PG *swanki- m. ‘misery sorrow; toil; temptation’ all turn out to display the
root structure SUoRT. Even if Rasmussen (1989: 164) claims the expected toga/fuga outcome of that
phonotactic constellation to be SU-RT rather than SUoRT, his claim seems to be supported by only one
example, viz. Goth. saurga f. ‘sorrow’ < PG *sw rg - < PIE *s rgh-éh2-. Admittedly, no arguments
exist that may unfailingly contradict the claim of Rasmussen. Nevertheless, when adding the statistics of
one form in favour of SU-RT as against three in favour of SUoRT as also the circumstance that SU-RT is
the only of the many constellations ending in -RT where zero grade rather than o-grade is expected, cf.
Rasmussen (1989: 162), we should at least consider alternative options for PIE *s rgh-éh2-. In fact,
Rasmussen (1989: 159) has done so already in that he admits that some nouns of the type PG *CC-eh2-
may reflect thematisations of old root nouns, especially when stressed on the root or when coexisting
with an attested root noun.
As for the remaining three forms, the first, viz. (75) PG *hlauti- m. ‘share lot’ might have been
created in analogy with other nominal o-grade derivatives of the same root, e.g. PG * la t - f.
‘sacrificial blood; lot’ or *hlauta- m. ‘share lot’. The second viz. PG (76) *laudi- m. ‘form
appearance’ has been demonstrated to be an old thematic stem only transferred into an i-stem, albeit
probably already in Proto-Indo-European times, due to its appearance as a second member of an

118
exocentric compound. The last one, viz. (78) PG *rauki- m. ‘smoke’ may be derived not from PG
*reukana- s.v. ‘smoke’ (originally ‘smell’) directly but from *raukiana- w.v. I ‘smoke’ if the latter is to
be analysed as a causative i.e. ‘make smell’ > ‘smoke’ rather than as a denominal verb.

Table 6: Expected radical ablaut grade in the toga/fuga system unknown

Derivational match with the ablaut grade of the stem of No derivational match with the ablaut grade of the stem
the preterite participle of a corresponding strong verb of the preterite participle of a corresponding strong verb

(5) PG *hwini- m. ‘whistling whiz’ (79) PG *saudi- m. ‘meat broth’

(8) PG *skridi- ~ *skriþi- m. ‘slide step’ (95) PG *staki- m. ‘mark scar’

(15) PG *stridi- m. ‘pace step’ (98) PG *awi- f. ‘ewe female sheep’

(16) PG *striki- m. ‘dot line’ (100) PG *duni- m. ‘noise’

(17) PG *swik(w)i- m. ‘delay postponement’ (112) PG *wurmi- f. ‘worm’

(18) PG *wliti- m. ‘face appearance form’

(19) PG *writi- m. ‘sign letter’

(20) PG *druzi- m. ‘fall’

(22) PG *gruzi- m. ‘fear dread’

(25) PG *hruzi- m. ‘fall death’

(26) PG *kuzi- m. ‘choice’

(28) PG *luzi- m. ‘loss’

(32) PG *sudi- m. ‘decoction extract’

(33) PG *supi- m. ‘drink’

(36) PG *bruni- ‘fire flame’

(37) PG *brusti- m. ‘break fracture’

(42) PG *ru(n)ni- m. ‘run course’

(48) PG *swuli- m. ‘swelling’

(57) PG *muni- m. ‘thought feeling’

(58) PG *numi- m. ‘taking’

(62) PG *agi- m. ‘fear’

(63) PG *aki- m. ‘ache pain’

(65) PG *skapi- m. ‘nature character kind’

119
(66) PG *slagi- m. ‘hit blow stroke’

(67) PG *stadi- m./f. ‘place town’

(68) PG *stapi- m. ‘step’

(70) PG *fangi- m. ‘catch’

(73) PG *walli- m. ‘source spring well’

Regarding roots of the structure reflected in the Proto-Germanic lexemes of table 6, we are able to
find evidence in Rasmussen (1989: 158-175) in favour of neither zero grade nor o-grade of the
toga/fuga type. For that reason, we shall not go into further details with these i-stem nouns, many of
them being clearly secondary, anyway; suffice it here to mention two, viz. the onomatopoetic (5)
PG *hwini- m. ‘whistling whiz’ and (67) PG *stadi- m./f. ‘place town’ being in reality, a ti-stem
rather than a primary i-stem.
(98) PG *awi- f. ‘ewe female sheep’ stands out from the rest of the forms listed in table 6 in that
it continues an o/e-ablauting, unanalysable i-stem PIE *h2 -i- ~ *h2 -i- ‘sheep’. Consequently
any systematic comparison between that and the remaining non-ablauting and deverbal i-stems
would seem futile.

4.2. I-stems resisting analysis within the toga/fuga system

Finally it should be noted that a set of i-stem nouns fall outside the scope of a toga/fuga analysis
seeing that they display radical full grade or lengthened grade, i.e. any other ablaut grade than o-
and zero grade. Two of these i-stems, viz. (60) PG *kwedi- m. ‘talk’ and (61) PG *wreki- m.
‘pursuit’ are synchronically analysable as verbal abstracts whose radical ablaut grades match the
stem of the preterite participle of a corresponding class V strong verb. We could theoretically say
the same of (71) PG * r - m. ‘shout’ but as pointed out by Hinderling (1967: 159) this i-stem
does not exist.
The remaining nouns cannot be matched with the ablaut grade of the stem of the preterite
participle of a corresponding strong verb, for which reason alternative explanations of their form
and structure must be sought. The i-stem nouns in question are (77) PG *leudi- m. ‘man’ (pl.
‘people’) (89) PG *sweli- m. ‘callosity callous skin’ (93) PG *kwemi- m. ‘eventus i.e. outcome
result’ (110) PG *w d - f. ’clothes clothing’ (105) PG *kw n - f. ‘wife’ (107) PG *segi- m.
‘victory’ (90) PG *weni- m. ‘friend’ (91) PG *w n - m. ‘hope expectation’ and PG (111) *wr g -
m. ‘complaint accusation’. For these examples of which some may be to be regarded as reanalyses

120
of old s-stems, some as i-stem verbal abstract with misinterpreted ablaut grade due to the
homophony of the infinitive and preterite participle stem of the Germanic strong verbs in classes V
and VI, some as vrddhi-derivations and yet some as derivations of old root nouns, cf. the
etymological notes to each of them in sections 2.2 and 2.3.

5. Conclusion
In order to separate archaisms from innovations, we must first define in every concrete case what
characterises an archaic form. In this article, I have defined the structure, form and function of
archaic, primary i-stems in Germanic.
Primary i-stem verbal abstracts are synchronically and productively formed from the same ablaut
grade as is found in the stem of the preterite participle of the corresponding strong verb. This new
system together with, i.a., the reanalysis of old s-stems as i-stems has given rise to a wide array of
possible Germanic ablaut grades in the root syllable, e.g. zero grades, o-grades, full grades and
lengthened grades. Close scrutiny has revealed, however, that the only archaic i-stems are those
displaying radical o- and zero grade.
Radical o- and zero grade of i-stems are also found in other Indo-European branches, as are the
two main functional categories of i-stems, viz. adjectival agent nouns on the one hand and action
nouns or verbal abstracts on the other. These two formal types, i.e. the zero grade type and the o-
grade type, were originally identical, the radical phonotactics constituting the determining factor for
when to expect zero grade and when to expect o-grade. Such a complementary distribution of o-
and zero-grade is reminiscent of what Rasmussen (1989: 158-175) suggests for the Proto-Indo-
European eh2-stems of the type Lat. toga/fuga of which these i-stems should probably be seen as a
variant, i.e. adjectival agent nouns of the structure PIE *CóC-i- or *C´C-i- and action nouns or
verbal abstracts of the structure PIE *CoC-í- or *CC-í-. Seeing that I have so far carried out a
systematic study of only the Germanic i-stems, it is too early to tell if we can extend our hypothesis
so as to also include i-stems in the remaining branches of Indo-European; further studies will
therefore have to elaborate on that matter.
When transferred to Germanic, the original distribution of o- and zero grade was abandoned in
favour of a new ablaut distribution dependent on the ablaut system of the strong verbs, and the old
system is only recognisable through an, albeit large, handful of i-stems displaying aberrant ablaut
grades in comparison with the strong verbs corresponding to them.

121
Article no. 3: The outcome of PIE *#Hi- and *#Hu- in
Germanic

It has been established with a great amount of certainty that PIE *#Hu- > Gr. ὐ-
and that PIE *#Hi- > Gr. ἰ-. It still remains to be demonstrated what happens in
other Western Indo-European branches, including Germanic. In this article 29, I reject
the statement by Ringe (1988: 433) that PIE *#Hu- > PG *#u-? and propose the
possibility of differentiated outcomes dependent on the timbre of the PIE laryngeal,
viz. that PIE *#h1i- > PG *#i- and PIE *#h1u- > PG *#u- as assumed by most
scholars, but that PIE *#h2i- and PIE *#h2u- might yield PG *#ai- and PG *#au-,
respectively. Furthermore, I tentatively propose that PG *#au- > PG *#u- when
followed by a labial consonant; a development partially paralleled in Greek and in
English.

1. PIE *#Hi- and *#Hu-: the classical view

In many grammars of Proto-Indo-European one may come across the statement that the outcome of
PIE *#Hi- and *#Hu- in the individual Indo-European languages with the sole exception of the
languages of the Anatolian branch is Post-PIE *#i- and *#u-, respectively, cf. e.g. Lehmann (1955:
32, 86-87), Lindeman (1987: 42-43), Beekes (1969: 128-129; 1988: 59 ff.) and – albeit somewhat
hesitantly – Cowgill (1965: 146-147). Quite recently, Clackson (2007: 57) uncritically adopts the
very same notion.
Given this vast, yet by no means exhaustive list of scholars arguing for the development of PIE
*#Hi- > Non-Anatolian-Post-PIE *#i- and of PIE *#Hu > Non-Anatolian-Post-PIE *#u-, any
mentioning of the idea that these two soundlaws would apply also for Germanic, i.e. that PIE *#Hi-
> PG *#i- and that PIE *#Hu > PG *#u-, would seem almost superfluous. A much-cited example of
the development of at least PIE *#h2u > PG *#u- is that of PG *ubila- adj. ‘evil bad’ with a related
form in Hitt. huwappa- adj. ‘evil ill bad’ derived from huwapp- ~ hupp- v. ‘be hostile towards do
evil against; throw (down) hurl’ cf. e.g. Watkins (1969: 30). By following the etymological
proposal of Watkins, Ringe (1988: 433) even tentatively judges the universal Germanic outcome of
PIE *#Hu- to be PG *#u-, cf. table 1.

29 This article has been submitted for publication in the journal Indogermanische Forschungen.

122
Table 1: nge’s s rve of lar ngeal develo ents n n t al os t on n t e Western Indo-European
languages (after Ringe 1988: 433)

PIE Greek Latin O[sco-]U[mbr.] P[roto-]C[eltic] P[roto-]G[erm.]

*N̥- a- *eN- (> iN-) aN- *aN- *uN-

*xN̥- aN- aN- aN- *aN- *uN-

*xwN̥- oN- *oN- (> uN-) ? *aN- ?

*xr̥- ar- *or- (> ur-) ? *ar- ?

*xu- au- au-? ? ? *u-?

Indeed, Watkins’ etymological proposal does seem to suggest a development of PIE *#Hu- > PG
*#u- as indicated in the table, but one seemingly safe example constitutes a comparatively meagre
body of evidence; hence probably Ringe’s application of a question mark in the table.
As such, it still remains to be seen if the communis opinio can be upheld or if the inclusion of
additional data will call for a new interpretation. By focusing particularly on the Germanic
outcomes of PIE *#Hi- and *#Hu-, this article aims at removing the insecurity exemplified by the
question mark in Ringe’s table.

2. The development of PIE *#Hi- and *#Hu- revisited

Though generally accepted, the notion of PIE *#Hi- and *#Hu- > Non-Anatolian-Post-PIE *#i- and
*#u does not gain support from everybody. Some scholars argue that, in some branches, notably
Greek and Italic, PIE *#Hi- and *#Hu- alternates with what some scholars choose to notate * ә -
and * ә - cf. e.g. table 1 for Ringe’s assumption of the development in Greek. The first to consider
this development was Pedersen (1909: 179): “Präidg. g im Wortanlaut vor einem w oder j + einem
unsilbischen Laut kann silbisch werden oder unsilbisch bleiben. Lat. augeō ’vermehre’ ir. uagim
’nähe’ S. 54 : skr. ugrá- ’gewaltig’ (idg. u- aus präidg. g ). Lat. ae-quus ’gleich’ : i-tem ’ebenfalls’
(zum Pronominalstamm *ei-, *ai-, *i- vgl. § 107).”
Hammerich (1948: 32) expresses similar ideas concerning the alternation between PIE *#Hu-
and * ә -: “[…] whereas H in Hu was preserved as consonantal in ’Anatolian’ primitively in
Aryan and Armenian, too, Hu generally became H in Greek, Italo-Celtic, Balto-Slavonic and –
perhaps – Albanese. Naturally, the au of several IE languages may sometimes be IE au and
sometimes IE H . But where the western languages have au- in forms corresponding to forms with

123
u- in the eastern languages, it is probable that we have western H - derived from IE Hu- preserved
as ḫ - in Hittite, as u- in the other eastern languages.”30
Heavily inspired by Lex Rix31 and slightly modified by Hyllested & Cohen (2007: 13), Peters
(1980: 5-125) was the one to finally uncover the situation in Greek, i.e. if PIE *#Hi- > Gr. ἰ- or
ἰ- and similarly if PIE *#Hu > Gr. ὑ- or ὐ-. Peters (1980: 72) concludes that the seven forms
αὐγή f. ‘sunlight’ αὐ ή f. ‘human voice speech’ αὔξω v. ‘make large increase; strengthen’ αὐχέω
v. ‘boast’ ἦρι/ἠέριος etc. adv. ’early’ ἰαύω etc. v. ‘sleep pass the night’ and εὖνις adj. ‘reft of
bereaved of’ may all have developed from PIE *#Hu- even though non-phonological explanations,
however unlikely, cannot be excluded. Against these seven examples, one single example, viz. Gr.
ὑφαίνω v. ‘weave’ seems to indicate a development of PIE *#Hu- > Gr. ὑ-, but by formulating the
rule that Pre-Gr. *#Vw- > #u- / _C[+lab], as a consequence of which Gr. ὑφαίνω should no longer
be regarded as a counterexample of the general development of PIE *#Hu > Gr. ὐ-, Hyllested &
Cohen (2007: 13) renders unnecessary any of Peters’ (1980: 114-125) attempts to explain why PIE
*h2ubh-n- would appear as Gr. ὑφαίνω rather than as †αὐφαίνω.
As for the development of PIE *#Hi-, Peters (1980: 113) claims that the comparison of the six
forms where Gr. ἰ- might be a result of PIE *#Hi- but where non-phonological explanations
might be equally attractive to the seven forms where Gr. ἰ- is seen as the result of PIE *#Hi- even
though non-phonological explanations, however unlikely, cannot be excluded clearly points in the
direction of PIE *#Hi- > Gr. ἰ- being the regular development. Contrarily, Joseph (1975: 322-323,
326-327) holds the rule of PIE *#HU- > Gr. # - to be valid for both glides. It is important to note,
however, that actual data from Greek plays a significantly minor role in Joseph’s article and that his
conclusions are based on theoretical and systemic considerations implying that the change PIE

30 With his citation of Friðþjófr Þórsteinsson’s Lausavísa no. 30 where Eyþjöfr (ON ey- < PG *au- with i-mutation)
and útsker should be seen as alliterating, Hammerich (1948: 33) even claims that Germanic, too, may have vocalised
PIE *H in initial position immediately preceding *u, cf. also his reconstruction of Goth. *austra- (Ostro-) < PIE
*H sro- < *Husro- or of Goth. aukan s.v. ‘increases grows’ < PIE *H g- (Hammerich 1948: 31).
31 PIE * H - > Gr. #VR-, cf. e.g. PIE *h2rĝ-ró- > Gr. ἀργός ‘shining bright’ (dissimilated from *ἀργρός) and PIE
*h3nbh-l-o- > Gr. ὀμφαλός ‘navel’ cf. Rix (1970: 84-102 and 84-85, 94-95 in particular).

124
*#HU- > # - is triggered by influence between various formal classes in the morphology rather
than by internal phonological pressures, cf. Joseph (1975: 327).32
According to Peters (1980: 113-114), the dissimilar developments of PIE *#Hi-and *#Hu- in
Greek are reminiscent of the situation found with the appearance of prothetic vowels, triggered by
laryngeals, in front of glides: The sequence of PIE * H - does develop a prothetic vowel in Greek
just as we would expect, cf. e.g. PIE *h2 es-s- > Gr. ἄεσα v. ‘spent (the night)’; PIE * H -, on the
other hand, does not.
With the sole exception of Latin, for which language Forssman (1982-1983: 291) in his review
of Peters (1980) tentatively suggests that PIE *#ә2u- > Lat. #au- based on the example of Lat.
a r ra f. ‘dawn’ (against which Schrijver (1991: 74-75)), there would seem to be no other Indo-
European languages in which PIE *#Hu- > * -. Neither are there any examples of PIE *#Hi- >
* -, not even in Greek if we choose to follow Peters (1980: 113) rather than Joseph (1975: 322-
323, 326-327). That Greek seems to take up a special position among the Indo-European languages
as regards the development of at least PIE *#Hu- is far from surprising in that Greek is one of only
three branches to display prothetic vowels developed from laryngeals in initial position followed by
a non-syllabic sound.33 As indicated above, yet contrary to what Hammerich (1948: 31) supposes,
the initial vowel of Gr. ὐ- should probably rather be seen as a prothetic vowel than as the
manifestation of a vocalised laryngeal, i.e. of PIE *ә.

3. Possible outcomes in Germanic

Turning now our attention towards Germanic, i.e. the branch for which Ringe (1988: 433) found it
necessary to apply a question mark in his table illustrating the development of PIE *#Hu- in a range
of Western Indo-European languages, we first need to set up a range of criteria in order for us to
determine what Germanic phonemes or combinations of phonemes in initial position could
theoretically reflect PIE *#Hi- and *#Hu-.

32 Equally or maybe even more important is the fact, mentioned by Joseph (1975: 323) himself, that the small amount
of data included in the article stems from an early draft of Peters (1980) possibly from a time prior to Peters’
completion of his data analysis.
33 The two other branches displaying vowel prothesis are Armenian and Phrygian, cf. e.g. Winter (1965: 100-101),
Klingenschmitt (1970: 80, 86), Polomé (1980: 27-28), Olsen (1985: 5-17; 1988-1989: 481-483) and Fortson (2004:
342, 402).

125
Based on the knowledge from other branches, it feels safe to assume that PG *#i- and *#u- are
possible outcomes, cf. e.g. Skt. k át v. ‘grows’ (< PIE *h2uk-s-é-). The special development of
PIE *#Hu- > Gr. ὐ- suggested by, e.g., Pedersen, Hammerich and Peters, cf. above, should also
be considered for Germanic. In Greek, according to Peters (1980: 7), the timbre of the vowel is
determined by the quality of the laryngeal, hence PIE *#h1u- > Gr. εὐ-, PIE *#h2u- > Gr. αὐ- and
PIE *#h3u- > Gr. οὐ-. If Germanic behaves in a way exactly parallel to Greek, the expected
outcomes could be PIE *#h1u- > PG *#eu-, PIE *#h2u- > PG *#au- and PIE *#h3u- > PG *#au-
with the vowels representing prothetic vowels even if Germanic, unlike Greek, is not normally
regarded as a language that develops prothetic vowels.
If, however, we start reflecting about the nature of prothetic vowels, we might be given a
rationale for any appearance of prothetic vowels in Germanic. In my view, prothetic vowels must
have arisen as a consequence of sandhi developments. For the sequence of PIE *#Hi- and *#Hu-,
four possible sandhi environments may occur, viz. PIE *-V#Hi/uV- (> *- ̄ -), *-V#Hi/uC- (>
*-Vi/uC-), *-C#Hi/uV- (> *- H ) and *-C#Hi/uC- (> *-CHi/uC-). In the former two
environments, we would expect the laryngeal to colour and, in the first example also lengthen, the
preceding vowel after which it would disappear in non-Anatolian-Post-PIE. The latter two
environments are of greater interest to us. In the first of these, the laryngeal would develop into a
true schwa which, however, has been preserved in this position as a so-called prothetic vowel only
in Greek, Armenian and Phrygian, cf. above. If we consider for the last sandhi environment that the
sequence PIE *CHi/u would develop a supporting vowel in order to ease the pronunciation in a way
parallel to Sievers’ Law for the cluster PIE * > * , i.e. in this case PIE *CHi/u >
* HH , we have an explanation for any development of PIE *#Hi- > PG *#Vi- and PIE *#Hu- >
PG *#Vu- as also for the identical development in Greek.
For the sake of systematic completion, we should also examine, though, whether either PIE
*#h1i- > PG *#ei- (> *#ī-), PIE *#h2i- > PG *#ai- and PIE *#h3i- > PG *#ai- or generally PIE *#Hi-
> PG *#ai- are possible developments, cf. the triple representation of PIE *h1/2/3 in Greek.
Consequently, in order to include all theoretically possible outcomes of PIE *#Hi- and *#Hu- in
Germanic, this study will examine all Germanic lexemes with the initial phonological combinations
of PG *#i-, *#ai-, * ī-, *#u-, *#au- and *#eu-.
It should be noted that not all lexemes with these initial combinations may be regarded as
reflecting PIE *#Hi- or *#Hu-. Alternative sources for PG *#ai-, * ī-, *#au- and *#eu are full and
o-grade forms of roots with the structure *H -. PG *#i- and *#u- may also reflect the zero grade

126
of PIE * - and *# -; PG *#i- and *#u- even have a third possible source, viz. as a raised variant
of PG *#e- (< PIE *#e-) preceding a nasal plus another consonant and as the supporting vowel in
the sequence PG *#uR- developed from PIE * H , respectively. In addition to these sources,
analogical reshapings, onomatopoeias and lexical borrowings should also be taken into account.
The material will therefore be divided into five sections:

1 Material with an initial laryngeal followed by PIE *(V)i or *(V)u

2 Material with PG *#i- and *#u- reflecting the zero grade of PIE *# C- and *# C-
3 Material with PG *#i- reflecting PIE *e /_NC
4 Material with PG *#u- reflecting the supporting vowel of PG *#uR- < PIE * -
5 Material from other sources (analogical reshapings, onomatopoeias, lexical borrowings)

The material of the first section will be further divided into subsections in accordance with the
quality of the laryngeal and the (syllabic or non-syllabic) glide.

3.1. Material with an initial laryngeal followed by PIE *(V)i or *(V)u

In the following Proto-Germanic lexemes, it is theoretically possible that PG *#i-, *#ai-, * ī-, *#u-,
*#au- and *#eu- reflect a PIE initial laryngeal followed either by a full or o-grade vowel and a glide
or by a syllabic glide, i.e. the structures PIE *He -, *Ho -, *HiC-, *He -, *Ho - and *HiC-.

3.1.1. Possibility of PIE *#h1i-

1 PG *aima- m. ‘smoke steam; smell’; see (14).

2 PG *aina- num. ‘one alone any’ > Goth. ains num. ‘id.’ ON einn num. ‘id.’ OSw. n, n
num. ‘id.’ ODa. een num. ‘id.’ OE n num. ‘id.’ OFris. n, n num. ‘id.’ OS n num.
‘id.’ OHG ein num. ‘id.’ etc. Initial PG *#ai- also in e.g. PG *ainahan- adj. ‘single’ >
Goth. ainaha adj. ‘id.’ ON einga adj. ‘id.’ (indecl.) OE nga, nga, nga adj. ‘id.’; PG
*ainaka- adj. ‘only special’ > ON einkum adj. ‘id.’ (dat.pl.); PG *a nakj n- f. ‘widow’ >
ON ekkja f. ‘id.’ OSw. ænkia f. ‘id.’ ODa. ænkiæ f. ‘id.’; and PG *ainak(a)la- adj.
‘standing alone’ > Goth. ainakls adj. ‘id.’. Extra-Germanic comparanda abound, e.g. Gr.
οἴνη f. ‘one (on dice)’ Lat. ūn s num. ‘one’ OIr. óen, óin num. ‘id.’ Lith. víenas num.
‘id.’ OPr. ains num. ‘id.’ etc.; all from PIE *o no- num./adj. ‘one alone’; with different
suffixation cf. e.g. Skt. éka- num. ‘one’ Mitanni-Indic aika- num. ‘id.’ Av. a va- num.
‘id.’ OPers. aiva- num. ‘one alone’ and Gr. οἶος adj. ‘alone lonely’. Often regarded as a

127
derivative from the pronominal stem PIE *(h1)i- ~ *(h1)e- ~ *(h1 e -, i.e. PIE *h1o -no-, cf.
e.g. IEW (2005: 286) and Bammesberger (1990: 227), but competing etymological
proposals exist, cf. e.g. Kroonen (2013: 11) who proposes affiliation of PG *aina- with the
root PIE *h2e - found in *h2o -u- ‘eternity lifetime’. Literature: Bammesberger (1990:
227), Bjorvand & Lindeman (2000: 186-187), Boutkan & Siebinga (2005: 87), Holthausen
(1974: 4), IEW (2005: 286), Kluge/Seebold (2002: 233), Kroonen (2013: 11), Lehmann
(1986: 16), Lloyd et al. (1998: 990-993), Orel (2003: 9), Philippa et al. (2003: 659), Sehrt
(1966: 95-97), De Vries (1962: 97).
3 PG *ainia- m. ‘juniper’. Represented in ON einir m. ‘id.’ LG n e ke m. ‘id.’. If an extra-
Germanic comparandum is represented in Hitt. eyan- n. ‘a certain evergreen tree yew(?)’
a reconstruction along the lines of PIE *h1 -n- ~ *h1 -n-, i.e. an acrostatic neuter n-stem,
might not be far-fetched, cf. Kroonen (2013: 12). Lat. ūn -perus) m. ‘juniper’ which is
traditionally compared to PG *ainia- (< PIE *h1o -n- o-), would thus need to continue PIE
*h1oi-n-i- with an enigmatic, initial Lat. i-, and the comparandum of Lat. ūn er s m.
‘juniper’ and Lat. iuncus m. ‘rush’ MIr. áin sb. ‘bulrush juncus effusus’ must thus be
abandoned. Lat. iuncus < PIE * o n-ko- may, however, be the folk etymological source for
the initial i- of Lat. ūn er s. Alternatively, LG n e ke and other West Germanic forms
may have been folk etymologically influenced by PG *aina- num. ‘one’ in which case the
traditional etymology of ON einir, LG n e ke etc. < PG *jainia- may be seen as valid in
spite of its exclusion of Hitt. eyan- as a comparandum. Literature: IEW (2005: 513),
Kroonen (2013: 12), Orel (2003: 205), De Vries (1962: 97).
4 PG *a s ja na- w.v. II ‘rush’. Represented in ON eisa w.w. II ‘rush dash’. Initial PG
*#ai- also in e.g. PG *a skr ja na- w.v. II ‘roar rage’ > ON eiskra w.v. II ‘id.’ if this
should not rather be reconstructed as PG *a d skr ja na- and compared to PG *aida- m.
‘pyre’ for which see (14), but the existence of Icel. ískra w.v. II ‘be furious from
excitement or pain’ (< PG *īskr ja na-), which cannot continue a form with initial PIE
*h2, clearly points in the direction of PG *a skr ja na- w.v. II ‘roar rage’ belonging here
since the root of PG *a s ja na- w.v. II ‘rush’ is normally reconstructed as PIE *h1eish2-
‘move rapidly’ cf. the extra-Germanic comparanda of, e.g., Skt. í at v. ‘sets in motion
sends’ Skt. á at v. ‘refreshes becomes strengthened’ rá- adj. ‘strong lively’ (< PIE
*h1 sә2-ró-), e á- adj. ‘quick’ Av. a a- m. ‘rage fury’ Gr. οἶστρος m. ‘rage’ ἱαρός adj.
‘strong lively’ (< PIE *h1 sә2-ró-), Lat. īra f. ‘anger rage’ (< PIE *h1e s 2-eh2-). The

128
actual Germanic form must be regarded as a denominal verb derived from PIE *h1 s 2-
eh2- ‘rage movement’ vel sim. or maybe rather PIE *h1ísh2-eh2- if we choose to accept
Rasmussen’s (1989: 172) claim that the PIE *-o- appearing in the verbal nouns of the toga-
type is always dropped when adjacent to, i.a., a laryngeal, cf. also the semantically closely
related fuga-type as well as the causative-iterative Skt. á at v. ‘refreshes becomes
strengthened’ (< PIE *h1ish2- e-) where identical conditions prevail regarding the
distribution of o- and zero grade. Literature: IEW (2005: 299-301), Kroonen (2013: 13-14),
LIV (2001: 234), De Vries (1974: 98).
5 PG *aiþa- m. ‘oath’. Represented in Goth. aiþs m. ‘id.’ ON eiðr m. ‘id.’ OSw. þer m.
‘id.’ ODa. t m. ‘id.’, OE t m. ‘id.’ OFris. t , d m. ‘id.’ OS t m. ‘id.’ and OHG
eid m. ‘id.’ etc. Initial PG *#ai- also in e.g. PG *aid(i)a- n. ‘isthmus’> ON eið, eiði n. ‘id.’
OSw. þ n. ‘id.’. Outside Germanic we find the same derivative only in Celtic, e.g. OIr.
oeth m. ‘oath’ whence the Germanic lexeme may have been borrowed or vice versa cf.
also Marstrander (1911: 205), Kluge/Seebold (2002: 230) and also Casaretto (2004: 452)
who, however, remains sceptical towards the idea of a lexical borrowing in either
direction. Together with PG *aiþa-, the Celtic lexeme points at PIE *H -to-. As for the
timbre of the laryngeal, all three options are available: If suggesting PIE *h1, we would
invoke affiliation of this derivative to the root PIE *h1e - ‘go’ based on the term ON
eiðgangr m. ‘oath-walk’. A reconstruction with PIE *h1 also finds support in the existence
of a form – also related to the meaning ‘oath’ – from Post-PIE *ei-to- that can only reflect
PIE *h1e -to-, viz. Umbr. eitipes (< eitom epens ‘they took an oath’); further cf. also Gr.
οἶτος m. ‘fate’ and OPhryg. oito- ‘id.’. In support of PIE *h2 and a root PIE *h2e -
‘important speech’ older literature lists Gr. αἴνημι, αἰνέω v. ‘praise’ cf. e.g. IEW (2005:
11), but this equation is rejected by the majority of more recent etymological dictionaries,
cf. e.g. Bjorvand & Lindeman (2000: 174-175). The last of the options at hand, viz. PIE
*h3 in a root PIE *h3e - ‘trust’ with cognates in Hitt. hai- v. ‘believe trust be convinced’
has been suggested by Puhvel (1991: 9-10), but as stated by Kloekhorst (2008: 267), most
attestations of this verb point to a stem Hitt. - (< PIE *h2/3eH-) rather than †hai-. As
such, PIE *h1 -to- would seem the most attractive candidate for PG *aiþa-. Regardless of
the timbre of the laryngeal, however, radical zero grade does not seem to be a possible
option, cf. OIr. oeth. Furthermore, contrary to the verbal adjectives of the type PIE * r-tó-
‘dead’ substantival to-formations, which formally resemble vrddhi-formations of the

129
verbal adjectives, are normally stressed on the radical vowel, cf. e.g. Brugmann &
Delbrück (1906: 27, 408-409). Literature: Bjorvand & Lindeman (2000: 174-175, 178),
Boutkan & Siebinga (2005: 29-30, 95), Casaretto (2004: 452-453), Holthausen (1974: 8),
IEW (2005: 295), Kluge/Seebold (2002: 230), Kroonen (2013: 8, 15), Lehmann (1986: 20-
21), NIL (2008: 222, 229), Orel (2003: 10), Philippa et al. (2003: 658), Sehrt (1966: 89),
De Vries (1962: 95).
6 PG *aiþma- m. ‘son-in-law’. Represented in OE t m. ‘son-in-law, brother-in-law’
OFris. t , t o , t e m. ‘son-in-law, father-in-law’ OHG eidum, eidam m. ‘son-in-
law’. ncertain etymology but three proposals are worthy of consideration. Firstly if PG
*aiþma- is to be understood as ‘son/father-in-oath’ comparison to PG *aiþa- m. ‘oath’ for
which see (5), is straightforward. Secondly, it can be compared to Av. a ta- m. ‘proper
share; punishment’ Gr. αἶσα, ἴσση f. ‘part share destiny’ Osc. aiteis f. ‘part’ (gen.sg.)
which are all derivatives of the root PIE *h2e - ‘give contribute’ cf. further Gr. αἴνυμαι v.
‘take’ i.e. ‘give to your self’. This makes sense from the semantic point of view that PG
*aiþma- is ‘he who has a share in the inheritance of the daughter’. Thirdly it may be
analysed as a derivative of the Lallwort root found in PG *a þīn-, *a þ n- f. ‘mother’ for
which see (96); however, it is questionable, in my view, if a term for a mother whose
acquaintance is not made until adult life could have developed from a nursery word for
mother. In all, the pedigree of this lexeme must be regarded as too uncertain in order for it
to serve as a basis for any assumptions regarding the radical ablaut grade and the timbre of
a possible initial laryngeal. Literature: Boutkan & Siebinga (2005: 29), Holthausen (1974:
8), Kluge/Seebold (2002: 230), Kroonen (2013: 15), Lloyd et al. (1998: 979-981).
7 PG *aiwa- m., aiw - f., *aiwi- f. ‘law’; see (27).
8 PG *i- pron. ‘he she it’. Represented in Goth. is; ita pron. ‘he; that’ ON es, er pron.
‘who’ OFris. -er pron. ‘he’ OS it pron. ‘that’ OHG ir, er; iz, ez pron. ‘he; it’ etc. With
initial PG * ī- in e.g. PG *ī- > Goth. ei conj. ‘that’ ON í (first member of temporal
denotations such as í gær ‘yesterday’ í dag ‘today’ etc.) OE ī -lca) adj. ‘same’ (< PG *ī-
līkan-). To be reconstructed as PIE *(h1)i- ~ *(h1)e- ~ *(h1 e -, cf. also Skt. ayám, idám, a-
etc. pron. ‘he etc.’ Av. a ә pron. ‘id.’ Gr. ἴν pron. ‘him’ Lat. is, ea, id pron. ‘he etc.’; as
for the varying representations of the root in the same paradigm also within Germanic, cf.
e.g. Goth. eis pron. (m.pl.) ‘they’ (< PIE *(h1 e -es). The assumption of an initial laryngeal
in this pronominal root is only structurally motivated, cf. e.g. Benveniste (1935: 148-149).

130
It can be stated with certainty, though, that if a laryngeal is present initially, it must be PIE
*h1, cf. again e.g. Goth. eis (< PIE *h1e -es) and Lat. ea pron. ‘she’ (< PIE *h1e -eh2); any
other laryngeal would have resulted in colouring of the radical vowel. Literature: Boutkan
& Siebinga (2005: 90), IEW (2005: 281-283), Kroonen (2013: 268), Lehmann (1986: 207-
208), Lloyd et al. (1998: 1092-1107), Orel (2003: 203), Sehrt (1966: 223-233), De Vries
(1962: 105).
9 PG *idi- f. ‘work’ > Burg. *iþs (in personal names such as Idbertus, Idwinus), ON ið f.
‘profession job’. Initial PG *#i- also in e.g. PG *ida- n. ‘constant moving quivering’ >
ON ið n. ‘id.’ and PG * d ja na- w.v. II ‘move around restlessly’ > Icel. iða w.v. II ‘id.’.
Outside Germanic, the denominal verb is found also in Gr. ἰτάω v. ‘go here and there’ Lat.
t v. ‘go’ MIr. ethaid v. ‘goes’. ON ið f. ‘profession job’ has a variant viz. ON íð f. ‘id.’
which, together with the extra-Germanic comparanda, points at PG *idi- etc. continuing a
PIE ti-stem *h1i-tí- to the root PIE *h1e - ‘go’ cf. also Skt. íti-, ití- f. ‘going walking’ or
even more correctly PIE *h1 -ti-/*h1 -t -, cf. Kroonen (2013: 269). Any connection of
this group of words to the root PIE *h2e dh- ‘burn’ i.e. PIE *h2idh-i- as alternatively
suggested by, e.g., De Vries (1962: 282-283) would seem futile in the light of ON íð which
can probably only continue PIE *(h1)ei- or *(H)iH-. Literature: IEW (2005: 294-295,
though not mentioning PG *idi-), Kroonen (2002: 269), NIL (2008: 221, though not
mentioning PG *idi-), Orel (2003: 203), De Vries (1962: 282-283).
10 PG *iliþ- m., lj - f. ‘footsole’. Represented in ON il f. ‘id.’ OE ill, ile m. ‘footsole hard
skin’ OFris. ili, ile m. ‘footsole’ MLG le, lde, lt n. ‘callus’ etc. Some scholars have
supposed that PG *īlan- m. ‘fishing net, weight, anchor etc.’ > ON íli, ili m. ‘id.’
Norw.(dial.) m. ile ‘id.’ is also related to this root, cf. e.g. Kroonen (2013: 269 with lit.),
contra which De Vries (1962: 284-285). Initial PG *#i- also in e.g. PG *ilkan- m.
‘footsole’ > ON ilki m. ‘id.’ which however has been explained by Hyllested (2008) as a
lexical borrowing from Saami. No satisfactory etymology. Attempts have been made at
connecting PG *iliþ-, * lj - with the root PIE *h1e - ‘go’ as well as with the formally
dissimilar Gr. ἴλια f. ‘female body-parts’ and Lat. īl a f. ‘belly’ which are both without
etymology (< PIE *īl e 2-?), cf. e.g. de Vaan (2008: 298). Literature: Boutkan & Siebinga
(2005: 191-192), Kluge/Seebold (2002: 232), Kroonen (2013: 269), Philippa et al. (2003:
458-459), De Vries (1962: 284-285).

131
11 PG *īl ana- w.v. I ‘rush hurry’. Represented in OS īl an w.v. I ‘strive; hurry’ OHG
īl l an, īlen w.v. I ‘id.’ etc. Initial PG *#ī- also in e.g. PG *īl - f. ‘hurry haste’ > OHG īla
f. ‘id.’. Often connected to the root PIE *h1e - ‘go’ i.e. as PIE *h1e -(e)lo- vel sim., cf.
IEW (2005: 296), or maybe as an l-derivative of an intensive to the root, i.e. PIE *h1e -
h1 -l e-, cf. Kroonen (2002: 269), but as rightly pointed out by Kluge/Seebold (2002: 232),
the original meaning of this verb i.e. ‘strive’ does not fit well with the semantics of *h1e -
‘go’. Kluge/Seebold therefore speculates if a zero grade PIE *ih2- to the root PIE * e 2- of
Skt. t v. ‘pursues revenges; pleads begs’ Gr. ζῆλος m. ‘eagerness’ would not be a
better candidate, especially in the light of the fact that a nearly identical duality of
meaning viz. ‘rush hurry’ ~ ‘pursue’ on the one hand and ‘plead beg’ ~ ‘strive’ on the
other, seems to be present in Skt. t ; this etymological proposal is rejected by Bjorvand
& Lindeman (2000: 430-431) on unspecified formal grounds. Literature: Bjorvand &
Lindeman (2000: 430-431), IEW (2005: 296), Kluge/Seebold (2002: 232), Kroonen (2013:
269), Philippa et al. (2005: 498-499), Sehrt (1966: 293).
12 PG *īsa- n. ‘ice’. Represented in Goth. iiz (name of the i-rune), ON íss m. ‘ice’ OE īs n.
‘id.’ OFris. īs n. ‘id.’ OS īs n. ‘id.’ OHG īs n. ‘id.’ etc. Often compared with Av. isu- adj.
‘cold icy frosty’ a xa- n. ‘cold’ but Av. isu- can only be derived from PIE * k̂ - (or
maybe * s-sk̂ -?); a PIE *isu- would result in Av. † -, cf. e.g. Hoffmann & Forssman
(1996: 102-104). Though semantically attractive, any attempt to unite Av. isu- (< PIE
*H k̂- or maybe *H s-sk̂-) with PG *īsa-, which must continue a PIE *HiH(-)s-o-,
*h1e -)s-o- vel sim. would at first hand seem futile. Judging from the prevailing neuter
gender in Germanic, however, I cannot help wondering if PG *īsa- could not, in fact,
continue a thematicised s-stem PIE *h1 -s-o- or maybe even *h1 -es-o- with the
phonological development of PIE *-e e- > PG *-ī- also recognised in, e.g., PIE *tr es
num. ‘three’ > PG *þrīz num. ‘id.’ and PIE * es pron. ‘we’ > PG *wīz pron. ‘id.’.34 Since
PIE *h1 -es-o- would probably yield PG †īza- rather than *īsa-, a reconstruction along the
lines of PG *h1 -s-o- is to be preferred. Av. isu- would thus not continue the s-stem but
rather be derived with an otherwise unknown (s)ku-suffix directly to the root PIE *h1e - in

34 Thematisation of s-stems is a trivial process in the Germanic languages, especially in Gothic and in North Germanic,
cf. e.g. Krahe (1967: 42-43), Thöny (2013: 82-84). In West Germanic, however, s-stems have been at least
peripherically preserved, cf. e.g. OE lemb n. ‘lamb’ (nom./acc.pl. lombur) and OHG lamb n. ‘id.’ (nom./acc.pl.
lembir < *lembiru < PG *la b z ).

132
the zero grade, i.e. < PIE *h1 - s k̂-u-. Such an analysis is, however, obstructed by Av.
a xa- whose Av. -x- can only be interpreted as PIIr. *-kh- (< PIE *-k(w)h1/2- or *-h1/2k(w)-).
Consequently, given the nonexistence of a nominal suffix PIE *-kHo-, the only possible
etymological analysis of Av. a xa- would include a root final laryngeal, i.e. PIE *h(1)e H-
vel sim. < PIE *h(1)e H-ko-/*h(1)o H-ko- (for the development of PIE *-h1/2-k- > *-kh- cf.
Olsen (1994: 274-275)), and thus exclude Av. isu-, which cannot continue a form with a
root final laryngeal, from the list of cognates unless we choose to follow de Vaan’s (2003:
246-250) suggestion that Av. ī may be shortened to i in some cases among which,
however, de Vaan (2003: 250) does not include that of Av. isu-. The addition of a root
final laryngeal offers no problems for the analysis of PG *īsa-; thus PG *īsa- < PIE
*h1e H-s-o-, *HiH-s-o- vel sim. but not PIE *h1e H-es-o- which would probably result in
PG †ejjes/za- > †ijjis/za-. Literature: Bjorvand & Lindeman (2000: 432-433), Holthausen
(1974: 189), IEW (2005: 301), Kluge/Seebold (2002: 236), Kroonen (2013: 271),
Lehmann (1986: 204), Orel (2003: 204), Philippa et al. (2005: 499), De Vries (1962: 287).
13 PG *īwa- m. ‘yew’. Represented in ON ýr m. ‘id.’ OE īw, ow m. ‘id.’ OHG īwa f. ‘id.’
(< PG *īw -). Initial PG *#ī- also in e.g. PG *ī a-/*īga- m. ‘id.’ > OE ī , o m. m. ‘id.’
OS īc as m. (pl.) ‘id.’ OHG īga, īgo (< PG *īg n -) m. ‘id.’. It is tempting to unite these
two lexemes as PG *ī wa- ~ *īgwa-, cf. e.g. Orel (2003: 203-204) and Holthausen (1974:
189) but Kroonen’s (2013: 271) etymological proposal of PG *īwa- < PIE *h1e H -o- or
*h1 H -o-, which finds support in extra-Germanic comparanda such as Lith. ievà f. ‘bird-
cherry’ Latv. ẽva f. ‘id.’ Ru. íva f. ‘willow’ (< PIE *h1 H- o-) and further Gr. ὄη, οἴη, ὄα
f. ‘elderberry tree mountain ash’ Lat. ūva f. ‘branch of grapes; raisins’ (< PIE *h1o H-
e 2-) and OIr. eó m.(?) ‘stem shaft yew-tree’ (< PIE *h1 eH- o-?), would render any
such attempt futile unless the root is reconstructed as PIE *h1e H- to which either of the
suffixes PIE *- o- or PIE *-ko-/*-k̂o- may be added. Consequently, it seems reasonable to
assume that PIE *#h1iH- > PG * ī-. Literature: Holthausen (1974: 189), IEW (2005: 297),
Kluge/Seebold (2002: 229), Kroonen (2013: 271), Orel (2003: 203-204), Philippa et al.
(2005: 498), De Vries (1962: 679).

3.1.2. Possibility of PIE *#h2i-

14 PG *aida- m. ‘pyre’. Maybe represented in Crim.Goth. (sched-)iit ‘light’ and with

certainty in OE d m. ‘(bon)fire pyre funeral pile’ OS d m. ‘firebrand’ OHG eit m.
‘fire oven’. Initial PG *#ai- also in e.g. PG *aidiana- w.v. I ‘burn (tr.) harden with fire’ >

133
OHG eiten w.v. I ‘id.’; PG *ai(d)la- n.(?) ‘flame’ > RN aila (cf. Grønvik 1996: 27), OE l
n. ‘flame’; PG *ai(d)liana- w.v. I ‘burn (tr.) ignite’ > OE lan w.v. I ‘id.’; PG *ai(d)lida-
m. ’fire’ > ON eldr m. ‘id.’ OSw. elder m. ‘id.’ ODa. eld m. ‘id.’ OE led m. ‘fire; fire
blight’, OS ld ‘id.’; PG *a d s n- f. ‘forge fireplace’ > ON eisa f. ‘embers’ MLG se f.
‘hearth forge fireplace’; PG *a st - f. ‘kiln’ > OE st f. ‘oven’ etc.; PG *ai(d)ma-35 m.
‘smoke steam; smell’ > ON eimr m. ‘reek vapor’ OSw. ember m. ‘id.’ OE m.
‘branding iron’; PG *aima- zj n- f. ‘embers’ > ON eimyrja f. ‘id.’ OE r e f. ‘id.’
OHG eimuria f. ‘pyre hot ash’; and maybe PG *a d skr ja na- w.v. II ‘roar rage’ > ON
eiskra w.v. II ‘id.’ if this should not rather be compared to PG *a s ja na- w.v. II ‘rush’
for which see (4). Extra-Germanic cognates include Skt. inddhé v. ‘ignite’ édha- m. ‘fuel’
Av. a s a- m. ‘firewood’ Gr. αἴθω v. ‘light up kindle’ αἶθος m. ‘fire embers’ ἰθαρός
adj. ‘clear bright shining’ Lat. aest s f. ‘hot season summer’ aed s f. ‘temple; room’
OIr. áed n. ‘heat fire’ Lith. íes f. ‘firewood’ etc.; all derived from the root PIE *h2e dh-
‘burn’. If also PG *idis-/*edis- f. ‘lady’ is to be affiliated with this root it would seem
obvious that PG *idis- < PIE *h2idh- in which case forms such as PG *aida-, *ai(d)la-,
*ai(d)ma- etc. can hardly represent the zero grade PIE *h2idh- but only the full grade PIE
*h2e dh- or the o-grade PIE *h2o dh-. As discussed under (33), however, PG *idis-/*edis-
probably has a different source. Literature: Bjorvand & Lindeman (2000: 429-430),
Holthausen (1974: 2-3, 10), IEW (2005: 11-12, 16-17), Kroonen (2013: 8, 11, 14),
Lehmann (1986: 298), Lloyd et al. (1998: 1024-1025), Orel (2003: 6-7, 10), Sehrt (1966:
94), De Vries (1962: 96, 98-100, 285).
15 PG *aigana- pp.v. ‘own possess have’. Represented in Goth. *aigan (pres. aih, aigum)
pp.v. ‘id.’ ON eiga (pres. á, eigum) pp.v. ‘id.’ OSw. g a pp.v. ‘id.’ ODa. g æ pp.v.
‘id.’ OE gan (pres. pp.v. ‘id.’ OFris. aga, ga pp.v. ‘id.’ OS gan (pres. g n pl.)
pp.v. ‘id.’ OHG eigan (pres. eigun pl.) pp.v. ‘id.’ etc. Initial PG *#ai- also in e.g. PG
*aigena- ~ *aigana- adj. ‘own’ (originally ptc. of PG *aigana- pp.v. ‘own possess have’)
> ON eiginn adj. ‘id.’ OE gen adj. ‘id.’ OFris. ein, eyn, egen adj. ‘id.’ OS gan adj.
‘id.’ OHG eigan adj. ‘id.’; PG *aihti- f. ‘belongings possessions property’ > Goth. aihts

35 ON ím n. ‘dust dirt’ Far. ím n. ‘soot’ which fit semantically well with this family of words are judged by Bjorvand
& Lindeman (2000: 432) to be secondary. Kroonen (2013: 11), however, prefers to reconstruct a second root *h1e -
on the basis of the comparandum of ON eimr and ím.

134
f. ‘id.’ ON ætt, átt f. ‘family race’ OE t f. ‘property power’ OHG t f. ‘id.’; PG
*a g n- f. ‘ownership property’ > ON eiga f. ‘id.’ OE ge f. ‘property’; and PG *aigni- f.
‘land property’ > ON eign f. ‘id.’. Extra-Germanic comparanda are found in Indo-Iranian
and Tocharian, cf. e.g. Skt. īśe v. ‘has at one’s disposal’ (< PIE *h2i-h2k̂-o ), Av. s v.
‘id.’ Toch. AB aik- v. ‘know recognise’ (< ‘have as one’s own be master of’). The timbre
of the laryngeal is determined by Toch. AB aik-, cf. Adams (1999: 101-102) who notes
that the consistent orthography in Tocharian of <aik>-, i.e. no occurrences of <eik>-,
points at a phonological sequence /āik-/ (< PIE *h2e k̂-) rather than /eik-/ (< PIE *h1o k̂- or
*h3e k̂-). In the light of the reduplicated perfects in Indo-Iranian and the preterite-presentic
character of the Germanic verb, it would seem obvious to reconstruct PG *aih as PIE *h2e-
h2 k̂-h2e and *aigum as PIE *h2e-h2 k̂-mé in which case this verb offers no case for a
development PIE *h2i- > PG *ai-. When focus is directed at the ti-stem PG *aihti-,
however, a development of PIE *h2 k̂-ti- > PG *aihti- cannot be excluded even though
radical full grade is equally possible, cf. e.g. Brugmann & Delbrück (1906: 429-438) and
particularly Bammesberger (1990: 144). Literature: Bjorvand & Lindeman (2000: 175-176,
179, 1081), Boutkan & Siebinga (2005: 10-11), Holthausen (1974: 2-3, 10), IEW (2005:
298-299), Kluge/Seebold (2002: 231), Kroonen (2013: 8), Lehmann (1986: 14), LIV
(2001: 223), Lloyd et al. (1998: 981-983), Orel (2003: 6), Philippa et al. (2003: 667-668),
Seebold (1970: 69-72), Sehrt (1966: 92-93), De Vries (1962: 95-96).
16 PG *aigena-, *aiginþ-? n. ‘shoot barb’. Represented in ON eigin n. ‘new sprout of corn’
NNorw eigind n. ‘grain germ barb’ etc. Initial PG *#ai- also in e.g. PG *aigla- m. ‘shoot’
> Sw.(dial.) egel, äjel m. ‘seed sprout’. These two formations are derived from a root PIE
*h2e k̂- ‘barb’ i.e. seemingly PIE *h2e k̂-ént- and *h2e k̂-(t)ló-, respectively; derivations
resembling the latter are found in, e.g., Gr. αἶκλοι m. (pl.) ‘corners of an arrow’ OPr.
ayculo f. ‘needle’ and Ru. iglá f. ‘id.’; further cf. Gr. αἰχμή f. ‘point of a spear spear’ Lith.
ẽ as, j ẽ as m. ‘spit broach’ < PIE *h2e k̂-smo/eh2-. In the light of the suffixal accent
in PIE *h2e k̂-ént- and *h2e k̂-(t)ló- (as revealed by the voiced Verner’s variant in
Germanic), radical zero grade would not be unexpected, cf. also Ru. iglá < PIE
*h2 k̂-tleh2-. On the contrary, seeing that PG *aiginþ-, if properly reconstructed as such, is
to be analysed as a participle of an athematic verb, radical zero grade is the standard, cf.
e.g. PIE *h1s-ént- ptc. ‘being’ and PG *aigla- is to be analysed as a concretised abstract

135
noun in which case radical zero grade and suffixal accent is actually to be expected.
Literature: IEW (2005: 15), Kroonen (2013: 9), Orel (2003: 6), De Vries (1962: 95-96).
17 PG *aik- f. ‘oak’. Represented in ON eik f. ‘id.’, OE c f. ‘id.’; as an ō- or i-stem in OFris.
k f. ‘id.’, OS k f. ‘id.’ (may also be a root noun); as an i-stem in OHG ei(c)h f. ‘id.’ (may
also be a root noun); and as a younger ō- or iō-stem in OHG eihhe, eihha f. ‘id.’. Initial PG
*#ai- also in e.g. PG *a kīna- adj. ‘oaken’ > ON eikinn adj. ‘id.’ OE cen, cen adj. ‘id.’
OFris. zen, tzen adj. ‘id.’ OHG e īn adj. ‘id.’. Often affiliated with the root PIE
*h2e ĝ- ‘shine’ cf. Gr. αἰγι- (e.g. in αἰγίλωψ m. ‘kind of oak’) Gr. αἴγειρος m. ‘poplar’
Lat. aesculus f. ‘durmast oak winter oak’ vel sim. (< *aigscolos), Lith. ąž olas, á ž ols,
á ž olas m. ‘oak’ etc.; further maybe ORu. jazvъ m. ‘badger’ Ru. jazь m. ‘carp’ OIr. áesc
sb. ‘concha clasendix’. If PG *aik- was originally a root noun rather than a vocalic stem as
supposed by, e.g., Griepentrog (1995: 24-32), I would a priori expect it to have shown
alteration between full or o-grade (PIE *h2e ĝ-/*h2o ĝ-) at the Proto-Indo-European stage
in the strong case forms and zero grade (PIE *h2 ĝ-) in the weak forms. However, with
only a few peripheral exceptions whose validity may all be debated, cf. Ph.D. article no. 1,
Germanic root nouns do not display ablaut. Rather, originally ablauting root nouns in
Proto-Indo-European eventually come to appear in a form in Germanic where three criteria
are fulfilled, viz. (1) that the root must contain at least one consonant in the syllable onset,
(2) that a vocalic element must be displayed in the root, and (3) that no more than one
consonant is allowed in the radical syllable coda. If, however, the root contains an original
PIE *a, that vowel will always be present in the root regardless of its phonotactic structure.
According to these criteria, PIE *h2e ĝ-/*h2o ĝ- ~ *h2 ĝ- would be generalised in the form
PIE *h2 ĝ- within Germanic. Given the validity of this assumption, it would seem that PIE
*h2 ĝ- > PG *aik-. Alternatively, of course, the presence of an original PIE *a in this root
cannot be excluded, i.e. PIE *a ĝ-; neither can the possibility of PG *aik- being a lexical
borrowing, cf. esp. Kroonen (2013: 9-10). Literature: Bjorvand & Lindeman (2000: 180),
Griepentrog (1995: 24-32), Holthausen (1974: 2), IEW (2005: 13), Kluge/Seebold (2002:
229-230), Kroonen (2013: 9-10), Lloyd et al. (1998: 974, 984-986), Orel (2003: 7),
Philippa et al. (2003: 669), De Vries (1974: 96).
18 PG *aikana- s.v. ‘make one’s own; assign allot’. Represented in Goth. (af-)aikan s.v.
’deny abjure’ OHG (in-)eihhan s.v. ‘claim’. Initial PG *#ai- also in e.g. PG
*a tr ja na- w.v. II ‘beg pray’ > Goth. aihtron w.v. II ‘id.’. Kroonen (2013: 10)

136
analyses this verb as PG *aikk- < *aig-n- < PIE *He k-n- by means of Kluge’s Law and
further regards it as derived from the verb PG *aigana- pp.v. ‘own possess’. Literature:
IEW (2005: 298-299), Kroonen (2013: 9-10), Lehmann (1986: 2, 14), Orel (2003: 6-7),
Seebold (1970: 72-73).
19 PG *aikiana- w.v. I ‘annoy pester’. Represented in Norw. eikja w.v. I ‘id.’. Initial PG
*#ai- also in e.g. PG *aikala- adj. ‘excited (by fear)’ > OE col, c l adj. ‘id.’ and PG
*aikena- ~ *aikana- adj./ptc. ‘wild furious’ > ON eikinn adj. ‘id.’ The verb PG *aikiana-
should be interpreted as a causative, i.e. < PIE *h2o g- e-, cf. Skt. ejayati v. ‘shakes’
derived from the verb PIE *h2 g-e- v. ‘move stir flutter’ > Skt. éjati v. ’id.’; further cf.
Gr. αἰγίς f. ‘rushing storm hurricane’ Lith. á k t s f. ‘passion glow’ OCS igra f. ‘game
fun’. Even if an original o-grade is seemingly secured for PG *aikiana- (< PIE
*h2o g- e-), this need not be the case for PG *aikena- ~ *aikana-: If it is, truly, a participle
of an unattested strong verb PG *aikana-, we would expect it to show the zero grade, i.e.
PG *aikena- ~ *aikana- < PIE *h2ig-enó- ~ *h2ig-onó-, unless PG *aikana- belongs to the
class of non-ablauting reduplicated verbs, cf. the pattern of Goth. aukan ‘increase’ (inf.)
aí-auk ‘increased’ (pret.1.sg.) aí-aukum ‘increased’ (pret.1.pl.) aukans ‘increased’ (ptc.)
(Hill 2009: 187-196; Krahe 1967: 105-106; Krause 1968: 235-236). As for this non-
ablauting reduplicated type in general, though, we could wonder what underlies the form
of the participle: That PIE *h2 g-e- > PG *aikana-, PIE *(h2)e-h2 g-h2e > PG *e-aik and
PIE *(h2)e-h2ig-mé > PG *e-aikum seems quite straightforward, but the expected form of
the participle should be PIE *h2ig-enó- ~ *h2ig-onó- in any case. Neither should it be
reduplicated PIE *(h2)e-h2ig-enó- ~ *(h2)e-h2ig-onó-; nor PIE *h2e g-enó- ~ *h2e g-onó-
with a radical full grade vowel. Unless analogical levelling has taken place, which is
indeed possible, it would therefore seem that PIE *h2ig-enó- ~ *h2ig-onó- > PG *aikena- ~
*aikana-. Literature: Holthausen (1974: 2), IEW (2005: 13-14), Kroonen (2013: 10), LIV
(2001: 222).
20 PG *aikwernan- m. ‘squirrel’. Represented in OE cweorna, cwern m. ‘id.’ eihhurno,
eihhorno m. ‘id.’. With initial PG * ī- in e.g. PG *īkwernan- m. ‘id.’ > ON íkorni m. ‘id.’.
Kroonen (2013: 10-11) manages to unite these two forms in an ablauting paradigm PG
*aikwernan- ~ *īk rnan- < PIE *h2e -h2 er-no- ~ *h2i-h2 r-no- with parallel or similar
formations in, e.g., Pers. varvarrah sb. ‘squirrel’ (< PIE *h2 er-h2 er-o-), Lat. vīverra f.
‘ferret’ (< PIE *h2 -h2 er-neh2-), OIr. íaru sb. ‘squirrel’ (< PIE *h2i-h2 er-(?)), W

137
gwywer sb. ‘id.’ (< PIE *h2 -h2 er-), Lith. vovere, v ver f. ‘id.’ (< PIE
*h2 e-h2 er- e 2-) etc. As such, there is no need for the etymological proposal frequently
advanced that the *aik- of PG *aikwernan- is related to the root PIE *h2e g- ‘move stir
etc.’ seen in PG *aikiana- w.v. I ‘annoy pester’ i.e. ‘squirrel’ < ‘swift little animal’ vel
sim. The mere existence of PG *īkwernan- also speaks against such an etymology: Its
radical vowel simply cannot be united with any root containing PIE *h2 unless, of course,
we choose to reconstruct a vrddhi-formation PIE *h2 g- > * k- > *eik- (by subsequent
application of Osthoff’s Law) > PG *īk- with non-colouring of the vowel, cf. Eichner
(1973: 72). Literature: Holthausen (1974: 2), IEW (2005: 13-14, 1166), Kluge/Seebold
(2002: 230), Kroonen (2013: 10-11), Lloyd et al. (1998: 974-976), Orel (2003: 7), Philippa
et al. (2003: 658), De Vries (1962: 284).
21 PG *aina- num. ‘one alone any’; see (2).
22 PG *airi adv. ‘early’. Represented in Goth. air adv. ‘id.’ ON ár adv. ‘id.’. Initial PG *#ai-
also in e.g. PG *airiz adv.comp. ‘before earlier’ > Goth. airis adv.comp. ‘id.’ ON ær
adv.comp. ‘id.’ OE r adv.comp. ‘previously; before’ OFris. r prep./conj. ‘id.’ OS r
adv.comp. ‘id.’ OHG r adv.comp. ‘id.’ and maybe ON áðr adv.comp. ‘id.’ (if to be
analysed as * ð r  * r r by dissimilation and not as *airþera-); further maybe PG
*aira-, *airu- m. ‘messenger’ > Goth. airus m. ‘id.’ ON árr m. ‘messenger servant’ OE
r m. ‘id.’ OS r m. ‘id.’; and PG *a r n ja na- w.v. II ‘be a messenger negotiate’ >
Goth. airinon w.v. II ‘id.’. PG *airi adv. ‘early’ is to be reconstructed as a locative PIE
*h2e -er-i of a heteroclitic stem PIE *h2e -r/n- ‘day’ with descendants in Av. a arә n.
‘day’ (gen.sg. a ąn) and Gr. ἦρι adv. ‘in the morning’. Literature: Bammesberger (1990:
66), Boutkan & Siebinga (2005: 90), Holthausen (1974: 6, 12), IEW (2005: 12),
Kluge/Seebold (2002: 228), Kroonen (2013: 12-13), Lehmann (1986: 19), Lloyd et al.
(1998: 1107-1109), NIL (2008: 258-259), Orel (2003: 9), Sehrt (1$966: 104-105), Philippa
et al. (2003: 662), Sehrt (1966: 104-105), De Vries (1962: 12, 14).
23 PG *a sk - f. ‘demand investigation’. Represented in OFris. ske f. ‘claim’ OHG eisca f.
‘question demand’. Initial PG *#ai- also in e.g. PG *a sk n- f. ‘question search
investigation’ > OE sce f. ‘id.’; PG *a sk ja na- w.v. II ‘demand inquire ask;
investigate, examine’ > OE sc an, cs an w.v. II ‘ask inquire’ OFris. sk a w.v. II
‘demand claim’ OS scon w.v. II ‘promote further’ OHG e sk n w.v. II ‘search look
for’; PG *a sk ng - f. ‘demand’ > OE sc ng f. ‘id.’ OHG eiskunga f. ‘id.’; and though

138
semantically weak, maybe PG *aiskaþla- n. ‘heart’ > ON eiskald n. ‘id.’. Extra-Germanic
comparandа are e.g. Skt. at v. ‘seeks’ iccháti v. ‘longs for’ cc - f. ‘wish demand’
Av. isaiti v. ‘longs for’ Arm. hayccem v. ‘beg’ aycc sb. ‘visit inspection’ Gr. ἵμερος m.
‘longing’ Lat. q aer v. ‘ask’ (< *ko-a s-e-), OIr. escaid v. ‘seeks’ Lith. íe ka v. ‘seek’
OCS iskati v. ‘search seek’ iska f. ‘demand’ kr. s’kát v. ‘search seek’ etc.; all from
the root PIE *h2e s- ‘search seek’. A ske-present seems to be widely distributed among the
Indo-European branches, albeit with two different radical ablaut grades, viz. PIE *h2 s-sk̂e-
> Skt. iccháti, Av. isaiti, OIr. escaid, OCS iskati, Ukr. s’kát 36 and PIE *h2e s-sk̂-e- > Arm.
hayccem, Lith. íe ka . Whether PG *aiskana-37 continues the expected ske-present form
PIE *h2 s-sk̂e- or the aberrant form PIE *h2e s-sk̂-e-, whose *h2e - may have been
introduced by means of influence from the perfect PIE *h2i/e-h2 s-e (> Skt. e a), cf.
Bjorvand & Lindeman (2000: 193), is virtually impossible to decide. Literature: Bjorvand
& Lindeman (2000: 193), Boutkan & Siebinga (2005: 27), Holthausen (1974: 7, 12), IEW
(2005: 16), Kluge/Seebold (2002: 403), Kroonen (2013: 13), LIV (2001: 260), Lloyd et al.
(1998: 1022-1024), Orel (2003: 9-10), Sehrt (1966: 108).
24 PG *aita- m. ’abscess ulcer’. Represented in OHG eiz m. ‘abscess boil’. Initial PG *#ai-
also in e.g. PG *a st n- f. ‘testicle’ > ON eista f. ‘id.’; PG *aitila- adj. ‘swollen’ > ON
Eitill m. (name of a sea king), EFris. eitel adj. ‘furious’ OHG eiz(z)ala f. ‘gallnut’; and PG
*aitra- n. ‘poison pus’ > ON eitr n. ‘venom poison’ OE tor, ter n. ‘poison’ OS ttar
n. ‘poison virus’ OHG eitar, eittar n. ‘poison pus’. Normally PG *aita- is reconstructed
as a verbal noun PIE *h2 d-o- derived from the root PIE *h2e d- ‘swell’ cf. also the extra-
Germanic comparanda of, e.g., Arm. aitnowm v. ’swell’ Gr. οἰ έω, οἰ άω v. ‘id.’ οἶ ος n.
‘swelling tumor’ Lat. aemidus adj. ‘swollen protuberant’ (< PIE *h2e d-(s)m-), Lith.
ìnkstas m. ‘testicle kidney’ (with nasal from ís os f. (pl.) ‘womb entrails interior’) and
OCS isto n. ‘testicle’ (pl. ‘kidneys’). In the light of Rasmussen’s (1989: 172) claim that the
PIE *-o- appearing in the verbal nouns of the toga-type is always dropped when adjacent
to, i.a., a laryngeal, PIE *h2íd-o- might be considered an option as well even though we

36 In the light of Ukr. s’kát , we must assume that PSl. *iskati < *jьskat < PIE *h2 s-sk̂-e-, cf. e.g. LIV (2001: 260); for
the opposite view that PSl. *iskati < *jīskat in spite of loss of the initial vowel in Ukrainian cf. Derksen (2003: 99-
100, 103).
37 No descendants from this verb are attested, though, but it must have existed since it serves as the basis for the
abstract noun PG *a sk -.

139
find Greek examples from PIE *h2o d-. Based on the obvious comparison of PG *a st n-
and Lith. ìnkstas, OCS isto n. ‘id.’ (probably < PIE *h2id-sth2-o-, cf. Kroonen (2013: 14)),
a zero grade may be considered here, too. Literature: Holthausen (1974: 8), IEW (2005:
774), Kluge/Seebold (2002: 237), Kroonen (2013: 14-15), LIV (2001: 258-259), Lloyd et
al. (1998: 1025-1029), Orel (2003: 10), De Vries (1962: 98).
25 PG *aiþa- m. ‘oath’; see (2).
26 PG *aiþma- m. ‘son-in-law’; see (6).
27 PG *aiwa- m., a w - f., *aiwi- f. ‘age eternity’. Represented in Goth aiws m. ‘id.’ OFris.
we f. ‘id.’ OHG wa f. ‘eternity’. Initial PG *#ai- also in e.g. PG *aiwin, *aiwan adv.
‘ever always’ (i.e. acc.sg. of PG *aiwi- f., *aiwa- m. ‘age eternity’) > Goth. (ni) aiw adv.
‘never’ ON æ, ei, ey adv. ‘always’ OE adv. ‘id.’ OS o adv. ‘id.’ OHG io adv.
‘always’ (n)io adv. ‘never’; PG *a wīn- f. ‘eternity’ > ON ævi f. ‘id.’ OHG wī f. ‘id.’;
and PG *aiwan- m. ‘id.’ > OHG wo m. ‘id.’. Extra-Germanic cognates abound, cf. e.g.
Skt. - n. ‘life lifetime vital power’ (gen.sg. o ‘health!’) ú- adj. ‘living vigorous
vital’ Av. - n. ‘life lifetime time’ (gen.sg. ao ), Gr. αἰών m./f. ‘(life)time long time
eternity’ αἰέν adv. ‘always’ (< PIE n-stem loc.sg. *h2e - -én ‘in eternity’) αἰές adv. ‘id.’
αἰεί adv. ‘id.’ (< PIE s-stem loc.sg. *h2e - -és-(i) ‘in eternity’) Lat. aevus m., aevum n.
‘period of time’. The Germanic as well as Latin forms are most likely to continue PIE
*h2e - -o-, i.e. a thematicised variant of the weak stem of the original, acrostatic neuter u-
stem PIE *h2 -u- ~ *h2 -u-, cf. e.g. Schindler (1975a: 7) and NIL (2008: 279), which was
later replaced by the more regular paradigm PIE *h2 -u- ~ *h2 -éu-, cf. e.g. the Indo-
Iranian forms. There is no compelling reason for assuming that PG *aiwa- m., a w - f.,
*aiwi- f. ‘age eternity’ continues a zero grade form PIE *h2 - -o- since descendants of
such a form are not attested in any other Indo-European branches. Literature:
Bammesberger (1990: 68), Bjorvand & Lindeman (2000: 1081-1082), Boutkan & Siebinga
(2005: 3-4), Casaretto (2004: 200-201), Holthausen (1974: 1, 8), IEW (2005: 17-18),
Kluge/Seedbold (2002: 264), Kroonen (2013: 16), Lehmann (1986: 22), Lloyd et al. (1998:
1175-1178), NIL (2008: 277-287), Orel (2003: 10-11), Philippa et al. (2003: 663), De
Vries (1962: 680, 682).
28 PG *aiwa- m., a w - f., *aiwi- f. ‘law’. Represented in OE , w f. ‘law religion
marriage’ OFris. -, wa, we, f. ‘law’ OS o, m. ‘id.’ OHG wa, a, o f. ‘law
right will contract’. Either to be reconstructed as PIE *h1o - -o- derived from the root PIE

140
*h1e - ‘go’ cf. Skt. éva- m. ‘course’ for an exact cognate or to be seen as identical to PG
*aiwa- m., a w - f., *aiwi- f. ‘age eternity’ for which see (27). In the latter case, which is
regarded the more likely alternative by most scholars, a both formal and semantic
connection can be established to Lat. ūs n. ‘law’ (gen.sg. ūr s, i.e. an s-stem) < PIE
*h2 -os ~ *h2 -es- which is probably in itself derived from the secondary weak stem of
PIE *h2 -u- ~ *h2 - - ‘age eternity’ cf. NIL (2008: 279), the semantic development of
PG* aiwa- etc. being one of ‘pertaining to eternity’ > ‘eternally valid’ > ‘law contract’.
Literature: Bammesberger (1990: 68), Bjorvand & Lindeman (2000: 1081-1082), Boutkan
& Siebinga (2005: 2-3), Holthausen (1974: 8), Kluge/Seebold (2002: 228), Kroonen (2013:
16), Lloyd et al. (1998: 1173-1175), NIL (2008: 277-287), Orel (2003: 10-11), Sehrt
(1966: 103).
29 PG *aiwiana- w.v. I ‘despise’. Represented in OE wan w.v. I ‘id.’. Initial PG *#ai- also
in e.g. PG *aiwiska- adj. ‘shameful’ > OE w sc adj. ‘shameless dishonoured’ MHG
eisch adj. ‘horrible’; PG *aiwiskia- n. ‘shame disgrace’ > Goth. aiwiski n. ‘id.’; PG
*a w sk - f. ‘dishonour disgrace offence’ > OE w sc; and PG *a w sk ja na- w.v. II
‘make ashamed treat shamefully’ > Goth. aiwiskon w.v. II ‘id.’. Etymology uncertain but
possibly to be compared to the root PIE *h2e gwh- ‘shame’ cf. Skt. an-ehás- adj. ‘flawless’
and maybe Gr. αἶσχος n. ‘shame’. We would expect the denominal verb PG *aiwiana- to
continue a PIE form with radical zero grade, i.e. *h2igwh- -, but the vocalism of the s-stem
PIE *h2e gwh-os ~ *h2e gwh-es- attested in Indo-Iranian and Greek may have influenced on
the verb for which reason it is virtually impossible to decide whether PG *aiwiana- and its
derivatives continue a radical full or zero grade. Literature: Holthausen (1974: 14),
Kroonen (2013: 16), Lehmann (1986: 21-22).
30 PG *aiza- n. ‘copper ore brass’. Represented in Goth. aiz n. ‘ore’ ON eir n. ‘brass
copper’ OE r, r n. ‘ore brass copper’ OS r n. ‘ore’ OHG r n. ‘id.’. To be
reconstructed as a thematicisation PIE *h2e -s-ó- or *h2e -es-ó-of the neuter s-stem PIE
*h2e -os ~ *h2e -es- ‘ore’ represented in e.g. Skt. áyas- n. ‘metal copper’ Av. aiiah- n.
‘id.’ Lat. aes ‘ore’ (gen.sg. aeris). Literature: Bammesberger (1990: 212), Bjorvand &
Lindeman (2000: 439), Holthausen (1974: 7, 11), IEW (2005: 15-16), Kluge/Seebold
(2002: 228), Kroonen (2013: 16-17), Lehmann (1986: 22), Lloyd et al. (1998: 1109-1110),
Orel (2003: 11), De Vries (1962: 97).

141
31 PG *a z - f. ‘peace clemency; respect benevolence’. Represented in ON eir f. ‘peace
clemency mercy’ OE r f. ‘honour dignity; kindness mercy’ OFris. re f. ‘honour
tribute’ OS ra f. ‘id.’ OHG ra f. ‘id.’. Initial PG *#ai- also in e.g. the denominal verbs
of PG *aiziana- w.v. I, *a z ja na- w.v. II ‘forgive; honour’ > ON eira w.v. II ‘spare
forgive’ OE r an w.v. II ‘respect’ OFris. aria w.v. II ‘id.’ OHG ren w.v. I, r n w.v. II
‘honour adore’ and PG *a st ja na- w.v. III ‘respect’ > Goth. aistan w.v. III ‘id.’ (< PIE
*h2e s-d(h3)-eh1- e-) i.e. literally ‘be in state of giving respect’. PG *a z - is normally
reconstructed as PIE *h2o s-eh2- to the root PIE *h2e s- ‘respect’ cf. also the extra-
Germanic comparanda of, e.g., Marruc. aisos m. ‘god’ (dat.pl.) mbr. esono- adj. ‘divine
sacred’ and maybe Gr. αἴ ομαι v. ‘respect’ (< PIE *h2 s-d(h3)-e-?). However, in the light
of Rasmussen’s (1989: 172) claim that the PIE *-o- appearing in the verbal nouns of the
toga-type is always dropped when adjacent to, i.a., a laryngeal, PIE *h2is-ó- might be
considered an option, as well. Literature: Bjorvand & Lindeman (2000: 1080-1081),
Boutkan & Siebinga (2005: 91), Holthausen (1974: 7), IEW (2005: 16), Kluge/Seebold
(2002: 228), Kroonen (2013: 17), Orel (2003: 11), Philippa et al. (2003: 661-662), Sehrt
(1966: 105), De Vries (1962: 97).
32 PG *idi- f. ‘work’; see (9).
33 PG *idis-/*edis- f. ‘lady’. Represented in OE ides f. ‘female lady’ OS idis, ides f. ‘wife’
OHG itis f. ‘(divine) woman’ and maybe if we choose to follow Grimm (1844 [1865]: 4-
5) and with him Kroonen (2012: 248-250), ON dís f. ‘woman girl; fairy nymph; goddess’
(seemingly < PG *dīs-i-). No further derivations; no immediate extra-Germanic cognates.
Further attempts at connecting this noun to material from other Indo-European branches
are semantically unsatisfactory and, if ON dís is included in the list of cognates, formally
improbable if not even impossible, cf. e.g. Bammesberger (2007: 83-85) who compares PG
*idis-/*edis- with Skt. édhas- n. ‘firewood’ and Gr. αἶθος m. ‘fire burning heat’ i.e. PG
*idis- < PIE *h2idh-es- vel sim. or Eichner & Nedoma’s (2000: 32-33) proposal of PG
*edis- < PIE *h1edh-es- to a root PIE *h1edh- also found in OHG etar m. ‘pale in a fence’.
Kroonen (2012: 248-250; 2013: 114-115) therefore sensibly suggests that PG *idis-/*edis-
is a lexical borrowing of unknown origin, cf. also the quite un-Indo-European “ablaut
pattern” of PG *idis- ~ *dīs-(i-). Literature: Bammesberger (2007: 81-89), Eichner &
Nedoma (2000: 30-34), Holthausen (1974: 185), Kroonen (2012: 248-251), Kroonen
(2013: 114-115), Sehrt (1966: 291), De Vries (1962: 77).

142
34 PG *īdala- adj. ‘void idle futile’; see (72).

3.1.3. Possibility of PIE *#h3i-

35 PG *a r - f. ‘oar’. Represented in ON ár, ǫr f. ‘id.’ OE r f. ‘id.’. Etymology uncertain

but probably comparable to Hitt. a- c. ‘drawbar’ Skt. ī - f. ‘pole of a wagon shaft’
Gr. οἴαξ m. ‘tiller handle of rudder helm’ Lith. íena f. ‘rod’ etc. i.e. PG *a r - < PIE
*h3o H-r-eh2- or *h3iH-r-eh2-. If Kroonen (2013: 13) is right in his speculations that the
Germanic and Baltic forms might reflect an old heteroclite PIE *h3o H-r ~ *h3 H-n-, the
former option, i.e. PG *a r - < PIE *h3o H-r-eh2-, seems most likely at first hand.
However, as the example of PIE *ud-r-ó- > Skt. udrá- m. ‘a kind of aquatic animal’
derived from the heteroclitic noun PIE * d-r ~ * d-n- ‘water’ demonstrates radical zero
grade is far from unexpected in this derivational type. Literature: Holthausen (1974: 7),
IEW (2005: 298), Kroonen (2013: 13), Orel (2003: 9), De Vries (1962: 12).
36 PG *aiþa- m. ‘oath’. See (6).

3.1.4. Possibility of PIE *#Hi- (i.e. undeterminable timbre of the laryngeal)

37 PG *aibra- adj. ‘harsh’. Represented in OE for adj. ‘vehement dire hateful’ OHG eipar,
eibar, eivar adj. ‘harsh rough’. Etymology unknown. If etymologically connected to PG
*ībra- m. ‘zeal eagerness’ > MHG īfer m. ‘id.’ and further to Lith. a br as m. ‘saliva
liquid from the mouth’ (< ‘bitter acrid taste’?) as assumed by a range of scholars but
rejected by probably equally many, cf. e.g. Heidermanns (1993: 96), a reconstruction with
initial PIE *h1 seems preferable; thus probably PIE *h1e bh-ró- > PG *ībra- and PIE
*h1o bh-ró- > PG *aibra- and Lith. aibr-. Whether either of the Proto-Germanic formations
could continue PIE *h1ibh-ró-, is difficult to decide; suffice it here to mention that radical
zero grade is expected with adjectives in PIE *-ro-. Albeit preferable, an etymological
analasyis dictating initial PIE *h1 is not the only option. Given the limited prevalence of
the root and derivative in question, virtually any initial laryngeal is possible. That
observation especially holds if PG *ībra- is not to be compared to PG *aibra-, if PG *ībra-
is to be analysed as a vrddhi formation, i.e. < PIE *H bh-ro-, or if the root also contains a
final laryngeal, i.e. PG *ībra- < PIE *HiHbh-ro- and PG *aibra- < PIE *Ho Hbh-ró- vel
sim. Literature: Heidermanns (1993: 95-96), Holthausen (1974: 2), IEW (2005: 11),
Kluge/Seebold (2002: 231), Kroonen (2013: 7-8), Lloyd et al. (1998: 969-970).
38 PG *īsa- m./n. ‘ice’; see (12).

143
3.1.5. Possibility of PIE *#h1u-

39 PG *auda- m./n. ‘riches wealth; fate destiny’; see (66).

40 PG *auja- n.? ‘luck fortune wealth’. Represented in Goth. awi-(liuþ) n. ‘thanks’ RN auja
n. ‘good fortune wellness’ ON ey sb. ‘luck fortune’. Initial PG *#ai- also in e.g. PG
*awidi- *awida-, *auþa-, *auþu-(?) adj. ‘easy comfortable’ > OE īeðe adj. ‘id.’ OS t
adj. ‘id.’ OHG d adj. ‘id.’ and PG *aw d , *a þ adv. ‘easily’ > ON auð-, OE aðe adv.
‘id.’ OS ðo adv. ‘id.’ OHG do adv. ‘id.’. Extra-Germanic comparanda are, e.g., Hitt.
iya(u)watta v. ‘recover’ Skt. ávati v. ‘helps supports’ vīt v. ‘has helped has supported’
(aor.), avitár- m. ‘patron benefactor’ Lat. v. ’support help’ OIr. (con)-oí v.
’protect’; all derived from a root PIE *h1e H- ‘help support’. According to LIV (2001:
243-244), Hitt. iya(u)watta secures the timbre of the initial laryngeal as PIE *h1; the
existence of the root final laryngeal is secured by the Indic se -forms. In Germanic, a PIE
*h1 H- o- would probably result in PG †ūja- rather than PG *auja- which, consequently,
must represent PIE *h1o - o- < *h1o H- o-. Based on the inclusion of Lat. ave v. ’am
well; am eager’ avidus adj. ‘desirous’ etc. as descendants of this root an alternative
etymology, cf. e.g. Kroonen (2013: 43), suggests that the root be reconstructed PIE *h2e -
‘enjoy consume’ in which case however Hitt. iya(u)watta can no longer belong here and
PG *auja- can reflect PIE *h2o - o- as well as PIE *h2e - o-. Given the validity of the
assumption that PIE *h2u- > PG *au-, a zero grade, i.e. PIE *h2 - o, could work as well.38
Literature: Heidermanns (1993: 110), Holthausen (1974: 87, 187), IEW (2005: 77-78),
Kroonen (2013: 43), Lehmann (1986: 52-53), LIV (2001: 243-244, 274), Orel (2003: 29-
30), Sehrt (1966: 424), De Vries (1962: 18-19, 106).
41 PG *aula(n)- m. ‘fool (tall) lanky fellow’. Represented in ON auli m. ‘id.’ NNorw. aul,
aule m. ‘angelica silvestris’. Extra-Germanic cognates are Hitt. auli- c. ‘tube-shaped organ
in the neck throat(?) windpipe(?)’ Gr. αὐλός m. ‘hollow tube pipe flute’ Lith. aũlas m.
‘leg of a boot; pipe of a mill’ a l s sb. ‘beehive’ av l s sb. ‘id.’ OPr. aulis sb. ‘shinbone’
etc.; all derived from the root PIE *h2e - with an *-l-suffix. If, as commonly assumed, ON
(hvann-)jóli m. ‘stalk (of angelica silvestris)’ and NNorw. jól sb. ‘angelica silvestris’

38 De Vries (1962: 18) and Lehmann (1986: 52) wish to separate PG *auja- from PG *awidi-, *awida-, *auþa-, *auþu-
adj. ‘easy comfortable’ etc. only to connect the latter with PG *auþia- adj. ‘remote empty desert desolate waste;
destroyed’ etc.

144
should be affiliated here, too, they must either be seen as the result of late, secondary
ablaut or be reconstructed as PG *eula(n)- < * la n - < PIE *h2 lo n -, i.e. a vrddhi
derivative to PIE *h2e -ló-; for the non-colouring of long vowels by laryngeal cf. Eichner
(1973: 72). A probably more obvious way of solving the problem of connecting PG
*eula(n)- to a root with an initial PIE *h2 is offered by Kimball (1994: 13-14). She states
that “[t]hese words cannot be derived from *h2 ul- and *h2 ul-, since the laryngeal is not
preserved in Hittite.” If then, we could reconstruct a root PIE *h1eu-, the o-grade could be
represented in PIE *h1o -lo/i- etc. > Hitt. auli-, Lith. a l s, PG *aula(n)- etc. and the full
grade in PIE *h1e -lo- > PG *eula(n)-.39 The vocalisms of Gr. αὐλός m. ‘hollow tube, pipe,
flute’ would then be interpreted as analogically influenced by the semantically similar
form Gr. καυλός m. ‘shaft stalk (of a plant) quill (of a feather)’ cf. also Güntert (1914:
154). A zero grade formation related to PG *aula(n)- and *eula(n)- might be attested
within Germanic, as well, viz. in PG *(haima- l j n- f. ‘sorrel’ > Icel. heimula, heimylja
f. ‘id.’ Norw. høymole, heimole f. ‘id.’ Swi. heimele f. ‘Good-King-Henry, chenopodium
bonus-henricus’ which must consequently reflect a PIE *h2 -l- eh2- or probably rather
*h1 -l- e 2-; cf. further Kolb (1957: 76) and with him Kroonen (2013: 42). Literature: IEW
(2005: 88-89), Kroonen (2013: 42), Orel (2003: 29), De Vries (1962: 292).
42 PG *auma- adj. ‘poor miserable’; see (67).
43 PG *eudra-, *ūdra- n. ‘udder’. Represented in ON júgr, júr n. ‘id.’ OE ūder n. ‘id.’
OFris. ūder, der n. ‘id.’ OHG ūtar, ūtaro, ūt r m. ‘id.’ etc. Extra-Germanic cognates are
Skt. ūd ar ~ ūd an- n. ‘id.’ (< PIE *h1uHdh-r/n-), Gr. οὖθαρ ~ οὖθατ- n. ’id.’ (< PIE
*h1o Hdh-r nt-), Lat. ūber n. ‘breast; udder’ (< PIE *h1e o Hdh-r-); the zero grade of the
root PIE *h1e Hdh- may also be what underlies Lith. ūdrúot v. ‘get milk; be pregnant’
Ru. úd t’ v. ‘ripen (of grain); swell up with liquid’. The even inner-Germanic alternation
between PG *eud- and *ūd- heavily points in the direction of PG *eud- continuing the full
grade PIE *h1e Hdh- and PG *ūd- continuing the zero grade PIE *h1uHdh-. Literature:
Bjorvand & Lindeman (2000: 444-445), Holthausen (1974: 375), IEW (2005: 347),
Kluge/Seebold (2002: 263), Kroonen (2013: 120), Orel (2003: 87), Philippa et al. (2009:
443), De Vries (1962: 294).

39 Admittedly, n-stems displaying radical ablaut are far from common in Germanic. It is therefore quite feasible that
the ablaut displayed in the Germanic examples is secondary.

145
44 PG *eup adv. ‘up upwаrds’. Represented in Goth. iup adv. ‘id.’. According to Johansson
(1891: 230-231), another phonotactically relevant derivative of this root may be PG
*e fn n- f. ‘crowd’ > Goth. iumjo f. ‘id.’ which is normally regarded as having no certain
etymology. PG *eup adv. ‘up upwаrds’ is to be regarded as an ablaut variant with full
grade (< PIE *h1e -) of PG *upp adv. ‘id.’. The alternation of PG *eup and *upp can only
point at PIE *h1 being the initial laryngeal, i.e. PIE *h1e - ~ PIE *h1up-. The gemination
of PG *upp – and originally maybe of PG *eup, too, if PG *eup- < *eupp- – may be caused
by the PG suffix *-n- associated with the system of directional adverbs; in this case by the
allative PG *-n(a) < PIE *-n-a/*-n-o, cf. Kroonen (2010: 371-373; 2013: 120-121).
Alternatively, however, the allatives PG *eup and *upp may both be regarded as
secondarily backformed from the locative PG *uppai adv. ‘up upon above’ cf. Kroonen
(2010: 374-376; 2013: 121), even if it remains unclear to the present author what could
have been the basis of the analogical backformation of PG *eup with full grade vocalism
from PG *uppai with zero grade vocalism. Literature: IEW (2005: 1107), Kroonen (2010:
367-379), Kroonen (2013: 120-121), Lehmann (1986: 208-209).
45 PG *eusizan- adj.comp. ‘better’. Represented in Goth. iusiza adj.comp. ‘id.’. Initial PG
*#eu- also in e.g. PG *e s l - f. ‘ease’ > Goth. iusila f. ‘id.’. Often connected to PIE
*h1 es-u- adj. ‘good’ cf. Skt. su- ‘well-‘ Gr. εὐ- ‘id.’. In the case of PG *eusizan-, a
Schwebe-ablauting full grade PIE *h1 s-is-(on-) is probably the most straightforward
reconstruction since radical, stressed full grade is expected in comparatives, cf. e.g
Brugmann & Delbrück (1906: 392-393, 547-562, esp. 557-558), as also revealed by the
general application of the unvoiced Verner variant in comparatives, cf. e.g. Goth. juhiza
adj.comp. ‘younger’ < PG *junhizan- to Goth. juggs adj. ‘young’ < PG *junga-. Still, the
PIE form of which PG *junhizan- is a descendant may have been stressed on the radical
vowel, but the root appears in the zero grade (PIE *h2 ú-h3nH-k-is-(on-)) rather than in the
full grade. Consequently, it does not seem impossible for PG *eusizan- to also continue a
zero grade, i.e. PIE *h1ús-is-(on-) rather than *h1 s-is-(on-). Literature: Heidermanns
(1993: 179), IEW (2005: 1174-1175), Lehmann (1986: 209), NIL (2008: 239-243).
46 PG *uba prep. ‘under; above’. Represented in Goth. uf prep. ‘under’ ON of prep. ‘over;
about’ OHG ob(a) prep. ‘above on over’. Initial PG *#u- also in e.g. PG *upp adv. ‘up
upwards’ > ON upp adv. ‘id.’ OE up adv. ‘id.’ OFris. up, op adv. ‘id.’ OS up adv. ‘id.’

146
OHG ūf adv. ‘id.’;40 PG * ban adv. ‘from above’ > ON ofan adv. ‘id.’ OE ufan, ufane
adv. ‘id.’ OFris. ova, uva adv. ‘id.’ OS ofan, ofana adv. ‘id.’ OHG obana adv. ‘id.’; PG
*ufuman- adj.comp. ‘highest upmost’ > Goth. aúhuma adj.comp. ‘higher’; PG *uber-
adv./prep. ‘over’ > Goth. ufar adv./prep. ‘id.’ ON yfir adv./prep. ‘id.’ OE ofer adv./prep.
‘id.’ OFris. over, ūr adv./prep. ‘id.’ OS obar adv./prep. ‘id.’ OHG ubar, ubari, ubir, ubiri
adv./prep. ‘id.’; PG * ft adv. ‘often’ > Goth. ufta adv. ‘id.’ ON oft adv. ‘id.’ OE oft adv.
‘id.’ OS ofto adv. ‘id.’ OHG ofto adv. ‘id.’; and PG *ubez- w -) f. ‘sth. tall; eaves’ >
Goth. ubizwa f. ‘portico’ ON ups, ux f. ‘eaves’ OE æfes, yfes f. ‘eaves brim brink’
OFris. se f. ‘gutters’ OHG obasa, obisa f. ‘portico entrance hall gallery’; further maybe
also PG *ubila- adj. ‘evil bad’ > Goth. ubils adj. ‘id.’ OE yvel adj. ‘id.’ OFris. evel adj.
‘id.’ OS ubil adj. ‘id.’ OHG ubil adj. ‘id.’; PG *ūba- adj. ‘ill-natured malicious’ > RN
ubaR, ON úfr adj. ‘unfriendly; bear, wolf’. Heidermanns (1993: 638) and Kroonen (2013:
557) regard the semantic connection between ‘under; above’ and ‘evil bad’ as one
travelling via ‘immense’ > ‘exceeding the boundaries overstepping a boundary’ i.e. ‘too
much; wrong bad’. Personally though I believe that the semantic starting point is not
‘above’ but rather ‘under’ in which case the semantic link between PG *uba and *ubila-
resembles that between G nieder adv. ‘down’ and niedrig adj. ‘mean’. Reconstructed as
PIE *h1upó prep. ‘under’ cf. also Skt. úpa prep. ‘towards together with under’ Αv.
prep. ‘towards with on in’ Gr. ὕπο, ὐπό prep. ‘under by’ Lat. s-ub prep. ‘under’ OIr. fo
prep. ‘under’ etc. and further maybe connected to Hitt. upzi v. ‘rises (of the sun)’. The
reconstruction with PIE *h1 is secured partly by Goth. iup adv. ‘up upwards’ partly by
Hitt. upzi without initial Hitt. h. If, however, Watkins (1969: 30) and Ringe (1988: 433) are
right in their assumption that PG *ubila- adj. ‘evil bad’ is rather related to Hitt. huwappa-
adj. ‘evil ill bad’ which is derived from huwapp- ~ hupp- v. ‘be hostile towards do evil
against; throw (down) hurl’ a reconstruction with PIE *h2 would seem more appropriate.
What Watkins fails to acknowledge, though, is that the original meaning of Hitt. huwappa-
was not ‘be hostile towards do evil against’ but rather ‘overthrow’ cf. Kloekhorst (2008:
369-371), for which reason Kroonen (2013: 557) chooses to reject the proposed connection
between Hitt. huwappa- and PG *ubila- and for which reason the reconstruction PG *uba,
*ubila- etc. < PIE *h1upó (with initial PIE *h1) can he upheld. If we want to maintain the

40 For the immediate etymology of this directional adverb see (44).

147
etymological connection between the Germanic and the Hittite forms, we can also offer an
alternative reconstruction, viz. that this group of words is actually to be reconstructed with
PIE *h3. In that case, the PIE *h3 o - reflected in Hitt. huwapp- would develop regularly
into PIE *h2 o -, cf. Cohen & Hyllested (2012: 57-58), from which the newly developed
PIE *h2 could be generalised into the zero grade stem of Hitt. hupp-. Literature: Bjorvand
& Lindeman (2000: 688-689), Boutkan & Siebinga (2005: 297-298, 421), Heidermanns
(1993: 636-638), Holthausen (1974: 240, 243, 376, 412), IEW (2005: 1105-1107),
Kluge/Seebold (2002: 76, 660, 938), Kroonen (2013: 120-121, 556-558), Lehmann (1986:
49-50, 208-209, 371-372), Orel (2003: 432-433), Philippa et al. (2007: 469, 484), Sehrt
(1966: 622-623, 630-631), De Vries (1962: 416-417, 632, 635, 676).
47 PG *unhta- adj. ‘accustomed’. Represented in Goth. (bi-)uhts adj. ‘id.’. Initial PG *#u-
also in e.g. PG *unhtia- n. ‘custom’ > Goth. (bi-)uhti n. ‘id.’. Extra-Germanic cognates are
found in, e.g., OIr. (do-)ucai v. ‘understand’ Lith. jùnkti v. ‘get used to’ (< PIE *h1u-n-k-)
and OPr. jaukint v. ‘exercise’ (< PIE *h1e k-n 2-); all ultimately derived from a root PIE
*h1e k- ‘get used to learn’ cf. further Skt. -ucyati v. ‘is used to takes pleasure in’ and
Arm. owsaw v. ‘learned’. PG *unhta- would thus need to be reconstructed PIE *h1ú-n-k-
to-. Literature: Kroonen (2013: 559), Lehmann (1986: 73-74), LIV (2001: 244-245).
48 PG * sl - f. ‘embers glowing ashes’. Represented in OE ysle, ysel f. ‘spark ash ember’
OHG usil(-far) adj. ‘ash-coloured’. Initial PG *#u- also in e.g. PG *uslan- m.
‘conflagration; embers’ > ON usli m. ‘id.’; PG * sj n- ~ *(aima- zj n- f. ‘embers’ > ON
ysja f. ‘fire’ eimyrja f. ‘embers’ OE r e f. ‘id.’ OHG eimuria f. ‘pyre hot ash’; and
maybe PG *ustr(j)a- adj. ‘very active’ > OHG ustar adj. ‘greedy’. Possible extra-Germanic
cognates are Skt. at v. ‘burns’ á- ptc. ‘burnt’ Gr. εὕω v. ‘singe’ Lat. ūr v. ‘burn’
ustus ptc. ‘burnt’; all derived from the root PIE *h1e s-. The Germanic forms undoubtedly
continue the zero grade of the root, i.e. PIE *h1us-. Literature: Bjorvand & Lindeman
(2000: 1004-1005), Holthausen (1974: 11, 413), IEW (2005: 347-348), Kroonen (2013:
562-563), LIV (2001: 245), Lloyd et al. (1998: 988-989), Orel (2003: 7, 436), De Vries
(1962: 96, 636).

3.1.6. Possibility of PIE *#h2u-

49 PG *au- prefix ‘away’. Represented in ON au-. Often reconstructed as PIE *au- (< PIE
*h2e -?) adv. ‘away’ also found in Skt. áva adv. ‘down (down) from’ Av. ava- ‘id.’ Gr.
αὔ-, Lat. au- ‘away (from)’ Lith. au- ‘id.’ OCS u- ‘id.’ etc. cf. De Vries (1962: 17-18).

148
The a-vocalism of Greek clearly points at the timbre of the laryngeal being PIE *h2, i.e.
PIE *h2e or *h2u. Peters (1980: 11) regards the option of PIE *h2u highly unlikely in that
the reflexes of this adverb in all branches contain a full vowel. IEW (2005: 55) advocates
for an alternative, etymological connection of ON au-, viz. to PIE *apu which is, in itself,
a variant of PIE *h2epó, *h2po prep. ‘of from away’ cf. e.g. Hitt. a adv. ‘behind
afterwards; back again further’ Skt. ápa prep. ‘away off’ Gr. ἄπο, ἀπό prep. ‘far away
away from’ Lat. ab prep. ‘from off’ PG *aba prep. ‘(away) from off’ (> Goth. af prep.
‘id.’ ON af prep. ‘id.’ OE æf, of prep. ‘id.’ OFris. of, af, ef prep. ‘id.’ OS af prep. ‘id.’
OHG aba, ab- prep. ‘id.’) OCS po- prep. ‘after on by at’ etc. Literature: Bjorvand &
Lindeman (2000: 51), Boutkan & Siebinga (2005: 292-293), Holthausen (1974: 9), IEW
(2005: 53-55, 72-73), Kluge/Seebold (2002: 3), Kroonen (2013: 1), Lehmann (1986: 2),
Lloyd & Springer (1988: 5-8), Orel (2003: 1), Philippa et al. (2003: 102), Sehrt (1966: 3),
De Vries (1962: 17-18).
50 PG *auda- m./n. ‘riches wealth; fate destiny’; see (66).
51 PG *auja- n.? ‘luck fortune wealth’; see (40).
52 PG *aukana- s.v. ‘increase augment’. Represented in Goth. aukan s.v. ‘id.’ ON auka s.v.
‘id.’ OFris. ka s.v. ‘id.’. Initial PG *#au- also in e.g. PG *aukan- m. ‘increase addition’
> ON auki m. ‘id.’ OE aca m. ‘id.’ OFris. ka m. ‘id.’; PG *a k ja na- w.v. II
‘increase augment’ > OE ac an w.v. II ‘id.’ OHG o c n w.v. II ‘id.’; and PG *aukiana-
w.v. I ‘id.’ > OSw. økia w.v. I ‘id.’ OS k an w.v. I ‘id.’. Being a strong verb, PG
*aukana- is expected to continue a PIE thematic present with radical full grade, viz. PIE
*h2 g-e-, which is actually attested also outside Germanic in Lith. áugti v. ‘grow’; further
cf. Tokh. auk- v. ‘grow increase’ Skt. ójas- n. ‘strength’ (< PIE *h2e g-es-), ugrá- adj.
‘strong powerful mighty’ (< PIE *h2ug-ró-), Lat. a ge v. ‘grow increase’ (< PIE
*h2o g- e-) and the extended root PIE *h2 eg-s- > *h2 ek-s- ‘id.’ in e.g. Skt. k át v.
‘grows’ Gr. αὔξω v. ‘grow’ ἀέξω v. ‘increase’ and PG *wahs(j)ana- s.v. ‘grow’ (> Goth.
wahsjan s.v. ‘id.’ ON vaxa, vexa s.v. ‘id.’ OE weahsan s.v. ‘id.’ OFris. waxa, wexa s.v.
‘id.’ OS wahsan s.v. ‘id.’ OHG wahsan s.v. ‘id.’) with further derivatives. One
circumstance deserves mentioning, though, viz. that PG *aukana- is not only a strong verb
but a strong, reduplicated verb. Consequently, we should consider how to properly
reconstruct the participle PG *aukena- ~ *aukana- ptc. ‘increased augmented’: as PIE
*h2e g-enó- ~ *h2e g-onó- or as *h2ug-enó- ~ *h2ug-onó-. Literature: Bjorvand &

149
Lindeman (2000: 1084-1085), Boutkan & Siebinga (2005: 13, 433), IEW (2005: 84-85),
Holthausen (1974: 83, 483), Kluge/Seebold (2002: 966), Kroonen (2013: 42, 566),
Lehmann (1986: 50, 387), LIV (2001: 274-275, 288-289), Orel (2003: 29, 439), Philippa et
al. (2009: 598), Sehrt (1966: 425-426, 633-634), Seebold (1970: 84-85), De Vries (1962:
19, 648).
53 PG *auk(e) adv./conj. ‘also; and’. Represented in Goth. auk conj./adv. ‘but also
furthermore’ ON auk, ok conj./adv. ‘also and’ OSw. och, ok conj./adv. ‘id.’ OE ac adv.
‘also however’ OFris. k adv. ‘also’ OS k adv. ‘id.’ OHG ouh adv. ‘id.’. Often seen as
identical with Gr. αὖ-γε adv. ‘again’ from the particle PIE *h2e - or *h2u-; for reflexes of
PIE *h2(e)u- cf. further Skt. u, utá conj./adv. ‘and also but’ Tokh. A -ok adv. ‘again’ B
-ok, -auk? adv. ‘id.’ (< PIE *h2e -ĝe, cf. also Adams (1999: 109)) and maybe Lat. autem,
aut conj. ‘but’. Seebold (1970: 84) alternatively regards PG *auk(e) as an imperative of PG
*aukana- v. ‘increase augment’ with a semantic development ‘increase!’ > ‘add (to that)!’
> ‘furthermore; also’. Literature: Boutkan & Siebinga (2005: 12-13), Holthausen (1974:
82), IEW (2005: 74), Lehmann (1986: 50), Kluge/Seebold (2002: 70), Orel (2003: 29),
Seebold (1970: 84), Sehrt (1966: 425), De Vries (1962: 19).
54 PG *aula(n)- m. ‘fool (tall) lanky fellow’; see (41).
55 PG *auma- adj. ‘poor miserable’; see (67).
56 PG *aura- m. ‘moisture water’. Represented in ON aurr m. ‘id.’ OE ar m. ‘wave sea’.
Probably identical to PG *aura- m. ‘(moist) earth soil’ > Goth. aura(-hjons)* f.
‘monuments tombs’ ON aurr m. ‘clay moist earth soil’ eyrr f. ‘shoal tongue of land
made up of sand and stone’ (< PG *a r - f.), OE ar, or m. ‘earth (of a grave)’. Outside
Germanic, PG *aura- is undoubtedly related to Gr. ἄναυρος adj. ‘without water’ (< PIE
*n-h2e ro- or maybe PIE *n-h2uro-, cf. Peters (1980: 55)), οὖρον n. ‘urine’ Lat. ūrīna f.
‘id.’. Any speculation whether Gr. -αυρος and PG *aura- continue a radical zero grade
form PIE *h2u-ro- or a morphologically unexpected full grade PIE *h2e -ro- is rendered
superfluous by the existence of PG *ūra- n. ‘soil(?)’ (> ON úr n. ‘moist drizzling rain;
metal slag soil containing iron’) of the probably related PG *ūr - m. ‘aurochs’ (> ON úrr
m. ‘id.’ ýrr f. ‘female auerochs’ (PG *ūr -), OE ūr m. ‘a kind of ox bison’ OHG ūro m.

150
‘aurochs’ (< PG *ūran- m.))41 and of the possibly related OIr. ūr, ū r sb. ‘earth clay’
(though cf. Kroonen (2013: 561)) which should all be reconstructed as PIE *h2uH-r- with a
root final laryngeal. Peters (1980: 55) mentions that at least Gr. -αυρος could still continue
a zero grade of the root in either of two ways: Either we could assume loss of laryngeal
due to the appearance of Gr. -αυρος as a second member of a compound, or we could
propose a Germanic (and Latin) sound change stating that PIE *u > PG/Lat. ū _r in
which case no laryngeal would be needed in order to explain the long initial vowel of PG
*ūra- etc. Literature: Bjorvand & Lindeman (2000: 1003-1004), Holthausen (1974: 85,
377), IEW (2005: 80), Kluge/Seebold (2002: 70-71), Kroonen (2013: 42, 561-562),
Lehmann (1986: 50, 379-380), Orel (2003: 29, 437), De Vries (1962: 20, 108, 635-636,
679).
57 PG *ausana- s.v. ‘scoop pour’. Represented in ON ausa s.v. ‘sprinkle pour’ MHG sen
s.v. ‘scoop out make empty’ œsen w.v. II ‘id.’ (< PG *ausiana-). Initial PG *#au- also in
e.g. PG *a s n- f. ‘bowl; ladle’ > ON ausa f. ‘ladle’ OE ease f. ‘bowl’. Possible extra-
Germanic cognates are Pal. hussinta v. ‘pour’ (3.pl.mid.), Gr. (ἐξ-)αύω v. ‘pour out’ and
Lat. hauri v. ‘draw scoop up’; both from PIE *h2 s- -, the latter with analogical full
grade and secondary h, though, cf. LIV (2001: 275). That the infinitive stem PG *ausana-
continues PIE *h2 s-e- is of only little debate, but with PG *ausana- being a strong,
reduplicated verb, we could wonder, though, how to properly reconstruct the participle PG
*auzena- ~ *auzana- ptc. ‘scooped poured’: as PIE *h2e s-enó- ~ *h2e s-onó- or as
*h2us-enó- ~ *h2us-onó- (in both cases with generalisation of the unvoiced Verner’s
variant). We could also consider if PG *ausiana- being as it is a continuation of a PIE ie-
present, could not be formally identical to Gr. (ἐξ-)αύω and Lat. a r < PIE *h2 s- -
even though a late, secondary formation of a PG *ausiana- on the basis of PG *ausana- is
indeed both possible and highly likely. Literature: Bjorvand & Lindeman (2000: 1090-
1091), Holthausen (1974: 86), IEW (2005: 90), Kroonen (2013: 43), LIV (2001: 275-276),
Orel (2003: 30), Philippa et al. (2005: 472), Seebold (1970: 85), De Vries (1962: 21).

41 Gąsiorowski (2012: 120) suggests an alternative etymology for PG *ūr - m. ‘aurochs’ viz. that it continues PIE
*h2us-ru- with a regular development of PG *-Vzr- > *- ̄ r-. The aurochs would then have to be regarded as ‘the red
one’ but as Gąsiorowski mentions himself the European male aurochs was black unlike the Skt. usrá- m. ‘bull’ and
sr - f. ‘(red) cow’ which are assumed by Gąsiorowski to be cognates of PG *ūr -.

151
58 PG *austera- adv. ‘east’. Represented in ON austr, aust- adv. ‘id.’ OE ast adv. ‘id.’
OFris. ster adv. ‘id.’ OS star, st adv. ‘id.’ OHG star, st adv. ‘id.’ etc. Initial PG
*#au- also in e.g. PG *a str n- f. ‘Easter’ > OE astre f. ‘spring goddess’ (pl. ‘Easter’)
OHG stara f. ‘Easter’. In related forms from other Indo-European branches, both PIE
*aus- (< *h2eus-) and PIE *us- (< PIE *h2us-) are found, cf. e.g. Skt. as- f. ‘dawn’ sr -
f. ‘id.’ Av. a - f. ‘id.’ Gr. ἕως, ἠώς, ἀώς, αὔως f. ‘id.’ Lat. a r ra f. ‘id.’ auster m.
‘south wind; south’ Lith. a r f. ‘dawn’ OCS za ustra ‘τὸ πρωΐ’ etc. Though criticised
by Forssman (1982-1983: 291), the notion of reconstructing a morphologically expected
radical full grade PIE *h2 s-os ~ *h2 s-es- for Gr. ἕως etc. and Lat. a r ra, cf. e.g.
Peters (1980: 31-32) and Schrijver (1991: 74-75), and an aberrant weak zero grade form
PIE *h2us-s- in order to explain Skt. as- and Av. a - seems to have gained general
support in the scholarly community. As for the formation with the contrastive or
comparative suffix PIE *-tero-, it is impossible to tell if a radical full or zero grade is
morphologically expected since such formations can be formed secondarily to virtually any
base as exemplarily illustrated by the reconstructions of the remaining cardinal points in
Germanic: PG *sunþera- adv. ‘south’ < PIE *sh2ún-tero- (zero grade, cf. also Kroonen
(2013: 492)), PG *nurþera- adv. ‘north’ < PIE *h1nr-tero- (zero grade, cf. also Kroonen
(2013: 393)), but PG *westera- adv. ‘west’ (no certain etymology; maybe < PIE * ekwsp-
tero-(?), cf. Kroonen (2013: 582-583), but undoubtedly radical full grade); cf. further
Brugmann & Delbrück (1906: 323-330, esp. 327) for the formation and nature of the suffix
PIE *-tero-. PG *a str n-, however, has probably been formed with the suffix PIE *-ro-
whose base normally appears in the zero grade, i.e. PIE *h2us-ró- (or, in this case, PIE
*h2us-reh2- + *-n), but in the light of the morphologically nearly identical forms Lith.
a r and OCS za ustra, the radical full or o-grade PIE *h2e o s-ró- (or PIE *h2e o s-reh2-
+ *-n) resembles a more probable alternative unless a development of PIE *h2u- > PBS
*au- can (also) be assumed. Literature: Bjorvand & Lindeman (2000: 1091-1092), Boutkan
& Siebinga (2005: 28), Holthausen (1974: 87), IEW (2005: 86-87), Kluge/Seebold (2002:
672), Kroonen (2013: 43), NIL (2008: 357-367), Orel (2003: 30), Philippa et al. (2007:
468), Sehrt (1966: 427), De Vries (1962: 21).
59 PG *auþia- adj. ‘remote empty desert desolate waste; destroyed’. Represented in Goth.
*auþ(ei)s adj. ‘barren desolate’ ON auðr adj. ‘desert empty’ OE adj. íeðe ‘desert
forlorn’ OHG d adj. ‘desert empty’. Initial PG *#au- also in e.g. PG *a þ - f. ‘desert’

152
> ON eyði f. ‘id.’ OHG d f. ‘id.’ and PG *auþiana- w.v. I ‘destroy’ > ON eyða w.v. I
‘waste destroy’ OE īðan w.v. I ‘lay waste destroy’ OHG (fir- den w.v. I ‘waste
desert’. Formally identical to Gr. αὔσιος adj. ‘empty vain’ and to be reconstructed as PIE
*a -t o-, i.e. application of the deadverbial suffix PIE *-t o- on the adverb/prefix PIE *au-
(< PIE *h2e -?) ‘away’ cf. Skt. áva adv. ‘down (down) from’ Av. ava- ‘id.’ Gr. αὔ-, Lat.
au- ‘away (from)’ Lith. au- ‘id.’ OCS u- ‘id.’ etc. Literature: Heidermanns (1993: 111),
Holthausen (1974: 187), IEW (2005: 73), Kluge/Seebold (2002: 662), Kroonen (2013: 43),
Lehmann (1986: 52-53), Orel (2003: 30), De Vries (1962: 18, 106-107).
60 PG *auzan- (~ *ausan-?) n. ‘ear’. Represented in Goth. auso n. ‘id.’ ON eyra n. ‘id.’ OE
are n. ‘id.’ OFris. re n. ‘id.’ OS ra n. ‘id.’ OHG ra n. ‘id.’. Normally reconstructed
as PIE *h2 s-(s)-on- ~ *h2e s-(s)-ón-, i.e. a bodypart-denoting n-stem derivative of the s-
stem PIE *h2 s-es- ‘ear’ cf. also Nussbaum (1986: 200-207, 210-212) though slightly
differently Lühr (2000: 291) and Schaffner (2001: 581). Cognates in the other Indo-
European branches are, e.g., Av. - n. ‘ear’ (only du. ), Gr. οὖς, ὦς αὖς n. ‘id.’ (gen.sg.
οὔατος), Lat. auris f. ‘id.’ OIr. áu, ó n. ‘id.’ Lith. ausìs f. ‘id.’ OCS uxo n. ‘id.’ etc. The
o-vocalism of Gr. οὖς etc., if not developed directly from PIE *h2 s-es-, could easily be
explained as caused by influence from semantically related Gr. ὤψ f. ‘eye’ < PIE *okw- <
*h3ekw-, though cf. also Szemerényi (1967: 65) and Peters (1980: 58-60), who alternatively
reconstruct the root as PIE *h3a s- ~ *h3us- with Gr. οὐσ- as the regular continuant of PIE
* h3us-. Whether PG *auzan- continues PIE *h2e s- or *h2us- is difficult to decide: If PG
*auzan- is derived from the PIE s-stem, i.e. PIE *h2e s-s-on-, only radical full grade is
expected with the sole exception of the dual form PIE *h2us-s-ih1 found in, e.g., Av.
‘(pair of) ears’ cf. e.g. Schindler (1975b: 259-260, 264). If, though less likely, it is derived
as an n-stem directly from the root, quantitative ablaut would be expected in which case
we would not be able to tell if the full grade form or the zero grade form had been
generalised or if the ablaut is, effectively, still present, i.e. if PG *ausan- < PIE *h2 s-on-
(or as a neuter rather PIE *h2 s-n) and PG *auzan- < PIE *h2us-én-. Literature: Bjorvand
& Lindeman (2000: 1087-1089), Boutkan & Siebinga (2005: 26), Casaretto (2004: 228),
Holthausen (1974: 85), IEW (2005: 785), Kluge/Seebold (2002: 664), Kroonen (2013: 44),
Lehmann (1986: 51), NIL (2008: 339-343), Orel (2003: 30-31), Philippa et al. (2007: 465),
Schaffner (2001: 575-584), Sehrt (1966: 426), De Vries (1962: 107-108).
61 PG *uba prep. ‘under; above’; see (46).

153
62 PG *ufna- m. ‘oven’. Represented in Goth. aúhns m. ‘id.’ ON ofn m. ‘id.’ OSw. ughn,
oghn, ofn, omn m. ‘id.’ ODa. ofn m. ‘id.’ OE ofen m. ‘id.’ OFris. oven m. ‘id.’ OHG
ofan, ovan m. ‘id.’ The velar consonants of Gothic and (Old) Swedish can be accounted for
as regularly developed from PG *f, cf. e.g. Hyllested (2012: 11) and Kroonen (2013: 557).
Consequently, the frequently cited comparanda of Skt. ukhá- m., k - f. ‘cooking pot’
and Lat. aulla, aula, auxilla f. ‘id.’ must be rejected so as for the true comparanda only to
include forms such as Hitt. huppar n. ‘bowl’42, Gr. ἱπνός, ἰπνός m. ‘oven’ Myc. i-po-no-
m. ‘dutch oven i.e. earthenware bowl used for baking on a hearth’ and OPr. wumpnis sb.
‘baking oven’ all of which are derived from a root PIE *h2e -/*h2up- with PG *ufna-
thus representing PIE *h2úp-no-. As pointed out by Kroonen (2013: 558), among others,
even these cognates generally vary too much and display too many irregularities for them
to have been regularly developed from one PIE root, for which reason the assumption is
cleverly presented that the word for ‘oven kiln’ etc. is a Wanderwort that has entered the
Western Indo-European languages individually. If the cognates from Sanskrit and Latin,
i.e. the forms with *-k(h)- rather than with *-p-, are included, as well, the likelihood of a
Wanderwort origin of the word for ‘oven kiln pot’ strongly increases. Literature:
Bjorvand & Lindeman (2000: 703-705), Holthausen (1974: 240), IEW (2005: 88),
Kluge/Seebold (2002: 663), Kroonen (2013: 557-558), Lehmann (2986: 49), Orel (2003:
433), De Vries (1962: 417).
63 PG *uhsan- m. ‘ox’. Represented in Goth. aúhsa m. ‘id.’ ON oxi, uxi (backformed from
pl. yxn) m. ‘id.’ OE oxa m. ‘id.’ OFris. oxa m. ‘id.’ (pl. ixen), OS ohso m. ‘id.’ OHG
ohso m. ‘id.’. Extra-Germanic cognates seem to be Skt. k án- ‘young bull’ Av. x an-
‘id.’ Toch. A ops, B okso m. ‘bull horned animal’ MIr. oss m. ‘stag cow’ MW ych m.
‘id.’ (pl. ychen). No clear etymology. Normally regarded as derived from PIE *ukw-s-on- to
the root PIE * egw- ‘wet moisten’ thus seen as ‘impregnator’; cf. further Skt. k át v.
‘spatters sprinkles moistens’ Gr. ὑγρός adj. ‘wet’ and Lat. uxor f. ‘wife’ (i.e. ‘the
impregnated one’). Semantically this etymology is not completely satisfactory: Zimmer
(1981: 84-91) points out that PIE *uk(w)son- can hardly be ‘a sprayer’ or ‘an impregnator’
since the Indo-Iranian cognates refer to a calf that has not yet procreated and the

42 Hitt. happena- ‘baking kiln fire-pit’ is rather to be compared with Gr. ὀπτός ‘baked’ < PIE * 3p-tó-, cf. Kloekhorst
(2008: 298).

154
Tocharian, Germanic and Celtic ones to a castrated bull, i.e. an ox. He further suggests that
PIE *ukson- be regarded as a Wanderwort. Kiehnle (1979: 118-119, 208-209) followed by
e.g. Pronk (2008: 1) alternatively suggests that at least the Indo-Iranian cognates reflect
PIE *h2uks-on- to the root PIE *h2 egs- ‘increase grow’ but for semantic reasons again
the remaining cognates would thus need to stand isolated. Literature: Bammesberger
(1990: 176), Bjorvand & Lindeman (2000: 683-684), Holthausen (1974: 243-244), IEW
(2005: 1118), Kluge/Seebold (2002: 662), Kroonen (2013: 558), Lehmann (1986: 49), NIL
(2008: 368-369), Orel (2003: 433-434), Philippa et al. (2007: 481-482), De Vries (1962:
422, 680), Zimmer (1981: 84-91).

3.1.7. Possibility of PIE *#h3u-

64 PG *auzan- (~ *ausan-?) n. ‘ear’; see (60).

65 PG *uba prep. ‘under; above’; see (46).

3.1.8. Possibility of PIE *#Hu- (i.e. undeterminable timbre of the laryngeal)

66 PG *auda- m./n. ‘riches wealth; fate destiny’. Represented in Goth. auda(-hafts) adj.
‘fortunate’ Burg. aud(s) m. ‘wealth’ ON auðr m. ‘fate destiny; wealth’ OE ad n.
‘possession riches property; happiness’ OS d n. ‘id’ MHG (klein- t n. ‘jewel gem’.
Initial PG *#au- also in e.g. PG *auda- adj. ‘rich’ > OE ad ‘id.’; PG *audaga-, *audiga-
adj. ‘rich; blessed’ > Goth. audags adj. ‘blessed’ ON auðigr adj. ‘rich’ OE ad g adj.
‘happy rich’ OS dag adj. ‘id.’ OHG tag adj. ‘id.’; and PG *audena- ~ audana- adj.
‘granted’ > ON auðinn adj. ‘granted ordained given’ OE aden adj. ‘id.’ OS dan adj.
‘id.’. Often compared with Lat. ūber adj. ‘rich fertile’ < PIE *Ho dh-ro-(?). Kroonen
(2013: 41) further adds Lith. áusti v. ‘weave’ and reconstructs PG *auda- as PIE *He -
dhh1-o- whose initial laryngeal must be identified as PIE *h2 if PIE *He - ‘weave’ cf. Skt.
ūv r v. ‘wove have woven’ (perf.3.pl.) (< PIE *H -H -r), is to be compared with the
extended root PIE *h2 ebh- ‘weave’ found in e.g. Gr. ὑφαίνω v. ‘weave’ (< PIE *h2ubh-n-
e-), cf. Hyllested & Cohen (2007: 13). We might consider, though, if the semantic
connection between PG *auda- m./n. ‘riches wealth; fate destiny’ on the one hand and
PG *auja- m. ‘luck fortune wealth’ and PG *awidi- *awida-, *auþa-, *auþu-(?) adj.
‘easy comfortable’ on the other is not close enough for them to be of common pedigree
i.e. for both to derive from either of the roots PIE *h1e H- ‘help support’ or PIE *h2e -
‘enjoy’ and if reconstructed PIE *h1e H-tro- > *h1e thro-, even Lat. ūber adj. ‘rich

155
fertile’ can easily be included; for the development PIE *-h1/2-tro- > *-thro- > Pre-Lat. *-
dhro- cf. Olsen (1988: 7-12), and for the formation of derivatives in PIE *-tlo-/*-tro- either
from agent nouns or, in sporadic cases, directly from heteroclitics, cf. Olsen (2010: 67).
The radical ablaut grades of the forms discussed here are difficult to establish due to the
great amount of uncertainty regarding their pedigree. However, if PG *audena- ~
*audana- adj. ‘granted’ is to be interpreted as a participle of an otherwise unattested
strong, reduplicated verb, cf. e.g. Orel (2003: 28) and Kroonen (2013: 41), we could
wonder how to properly reconstruct it: as PIE *He H -t/dh-enó-(?) ~ *He H -t/dh-onó-(?)
or as *Hu(H)-t/dh-enó-(?) ~ *Hu(H)-t/dh-onó-(?). Literature: Bammesberger (1990: 54),
Holthausen (1974: 83), IEW (2005: 76), Kroonen (2013: 40-41), Lehmann (1986: 47),
Orel (2003: 28), De Vries (1962: 18).
67 PG *auma- adj. ‘poor miserable’. Represented in ON aumr adj. ‘id.’ OSw. ömber adj. ‘id’
etc. Initial PG *#au- also in e.g. PG *a n- f. ‘misery’ > ON auma f. ‘id.’ and PG
*a k ja na- w.v. II ‘commiserate feel pity for’ > ON aumka w.v. II ‘id.’. Except for the
neat comparandum of Toch. B aume n. ‘misery’ cf. Adams (1999: 132) this lexeme has
no satisfactory etymology. Attempts have been made, though, to connect this adjective
with Gr. εὖνις adj. ‘empty’ (remodelled from *ῡνις < PIE *h1uh2-ni-, cf. Peters (1980: 51-
52)) to the root PIE *h1 e 2- ‘empty’ cf. also Skt. ūná- adj. ‘lacking missing’ Lat. v n s
adj. ‘hollow devoid’ PG *wana- adj. ‘lacking missing’ (> Goth. wans adj. ‘id.’ ON vanr
adj. ‘id.’ OE wan adj. ‘id.’ OS wan adj. ‘id.’ OHG wan adj. ‘id.’). Others prefer a
connection of PG *auma- adj. ‘poor miserable’ to PIE *au- (h2e -?) adv. ‘away’ found in
PG *auþia- adj. ‘remote empty desert desolate waste; destroyed’. A third etymological
proposal originally offered by Noreen (1923: 169) and most recently reintroduced by
Kroonen (2013: 35) regards PG *auma- adj. ‘poor miserable’ as dissimilated from PG
*arma- adj. ‘id.’ > Goth. arms adj. ‘id.’ ON armr adj. ‘id.’ OE earm adj. ‘id.’ OFris. erm
adj. ‘id.’ OS arm adj. ‘id.’ OHG aram adj. ‘id.’ etc. cf. the classical etymology suggested
by Johansson (1891: 223-224) that PG *arma- < *arbma- < PIE *orbh-mo- < *orbh-no-
‘orphan’ < PIE *h3orbh- also seen in Gr. ὀρφανός adj. ‘orphan; bereaved bereft’ and Lat.

156
orbus adj. ‘bereaved bereft’.43 Given the very limited distribution of PG *auma- adj.
‘poor miserable’ within the Germanic realm this etymological proposal with its inclusion
of irregular dissimilation presents itself as quite attractive. Literature: Heidermanns (1993:
109-110), Kroonen (2013: 35, further also 32-33), Orel (2003: 29), De Vries (1962: 19,
107).
68 PG * j ja na- w.v. II ‘sound’. Represented in Goth. aúhjon w.v. II ‘id.’ Icel. ýja w.v. II
‘remind drop a hint’. Possibly related to Latv. aũka f. ‘gale’ SCr. uka f. ‘shouting’; both
from PIE *Ho k-eh2-. Alternatively PG * j ja na- may be analysed as a zero grade
derivative of the verb PG *wahana- s.v. ‘remark’ to the root PIE * ekw- ‘speak’ or simply
as onomatopoetic. Literature: Kroonen (2013: 558), Lehmann (1986: 48-49).
69 PG *unþi-, nþ - f. ‘wave’. Represented in ON unnr, uðr f. ‘id.’ OE þ f. ‘id.’ OS ūt a
f. ‘id.’ OHG undea f. ‘flood wave’. In my view this lexeme may best be compared to Skt.
avatá- m. ‘well’ ava á- m. ‘pit; cavity’; thus PG *unþi-, * nþ - < PIE *(H)ú-nt-i- and
Skt. avatá- < PIE * H e -nt-ó-(?). Kroonen alternatively suggests a reconstruction PIE
*h2n-ti(h2) in that he compares the Germanic forms to Hitt. n ~ hananzi v. ‘draw
(liquids)’ and Gr. ἄντλος m. ’hold of a ship; bilge-water flood’. Far less convincing is the
etymological proposal mentioned by IEW (2005: 80) and De Vries (1962: 635) that PG
*unþi-, nþ - be a zero grade formation of PIE * et- ‘wet’ in itself a parallel root to PIE
* ed- ‘water’ cf. also Lat. unda f. ‘wave’. Literature: Holthausen (1974: 413), IEW (2005:
78-80, esp. 80), Kroonen (2013: 560-561), Orel (2003: 435), Sehrt (1966: 623), De Vries
(1962: 635).
70 PG *ūt adv. ‘out’. Represented in Goth. ūt adv. ‘id.’ OE ūt adv. ‘id.’ OFris. ūt adv. ‘id.’
OS ūt adv. ‘id.’ OHG ūz adv. ‘id.’. A variant of PG *ūt appears as PG *uz (< *ut-s +
C[+voice]) adv./prep. ‘id.’ > Goth. us, ur- adv./prep. ‘id.’ ON ór, or-, ør- ‘id.’ OE or- ‘id.’
OFris. or-, ur- ‘id.’ os. ur-, or- ‘id.’ OHG ur-, ar-, ir- ‘id.’. Extra-Germanic comparanda
include Skt. úd-, út- ‘up upwards’ Gr. ὑ-, Lat. ūs-(que) ‘continously, incessantly’, Lith.
ž- ‘up upwards’ OCS vъz-, vъs- ‘id.’ etc.; all from PIE *(H)ud adv./prep. ‘up, upwards;
out outwards’. Lengthening of the vowel in PG *ūt is probably caused by the

43 Meillet (1898: 280) has proposed an alternative etymology for PG *arma- adj. ‘poor miserable’ viz. that PG
*arma- < PIE *(h1)or-mo- adj. ‘weak’ also seen in Arm. ołor sb. ‘mercy’ adj. ‘piteous’ (dissimilated from *or-
orm) and in Hitt. erman-, arman- n. ‘sickness illness’.

157
monosyllabicity of the word. Literature: Bammesberger (1990: 199), Boutkan & Siebinga
(2005: 422, 425), Holthausen (1974: 241, 377), IEW (2005: 1103-1104), Kluge/Seebold
(2002: 74, 251), Kroonen (2013: 562), Lehmann (1986: 380, 384), Orel (2003: 436-437),
Philippa et al. (2007: 465-466), Philippa et al. (2009: 444), Sehrt (1966: 631-633), De
Vries (1962: 419, 635-636, 682).

3.2. Material with PG *#i- and *#u- reflecting the zero grade of PIE *# C- and *# C-

In the following Proto-Germanic lexemes, PG *#i- and *#u- should be regarded as the zero grade of
PIE * C- and *# C-.
71 PG *irha- m. ‘he-goat’. From PIE *irk̂-o-, i.e. a zero grade formation of PIE * ork̂-‘kind of
deer’ vel sim. seen in Gr.(Hes.) ἴορκες f.(?) ‘kind of deer roe gazelle’.
72 PG īdala- adj. ‘void idle futile’. From PIE *ih1-tlo-(?) to the root seen in W ial f.
‘brightening’ and Ru. jályj, jálovyj adj. ‘infertile unused (of land)’. An alternative
etymology suggests PIE *h2iHdh- to PIE *h2e dh- ‘burn’ which is though not known to
have had any word-internal laryngeal.
73 PG *īl ana- w.v. I ‘rush hurry’; see (11).
74 PG *-ud ‘year’ (only in PG *fer-ud ‘last year’). From PIE *ut-, i.e. formed to the zero
grade of the root PIE * et- ‘year’.
75 PG * j na-, * j jana- w.v. II ‘sound’; see (103).
76 PG *uhsan- m. ‘ox’; see (63).
77 PG *unþi- ~ nþ - f. ‘wave’; see (69).
78 PG *urzan- m. ‘black grouse capercaillie’. Probably from PIE *urs-on- ‘male animal’ i.e.
a zero grade of PIE * ers-on- seen in, e.g., Skt. sab á- adj. ‘manly powerful’ m. ‘bull
chief’ ( *v sab á- by reanalysis) and Gr. ἄρσην adj. ‘male’; cf. further Pronk (2008).
79 PG *utra- m. ‘otter’. From PIE *ud-ro-, i.e. formed to the zero grade of the root PIE * ed-
‘water’.

3.3. Material with PG *#i- reflecting PIE *e /_NC

In the following Proto-Germanic lexemes, PG *#i- should be regarded as a variant of PG *#e- in

front of a nasal followed by another consonant.
80 PG *in prep. ‘in’ *instar- n. ‘intestinal fat’ *inþera- n. ‘entrails’ etc. (< PIE *h1en- ‘in’).
81 PG *inkan- m. ‘pain torment’. From PIE *h1eng- ‘torture press’ vel sim.

158
82 PG *inkwa(n)- m. ‘lump’. From PIE *engw-, i.e. a secondary full grade of the root PIE
*negw- ‘lump wound’ vel sim. also found in, e.g., Lat. inguen m. ‘swelling on the groin;
groin’.

3.4. Material with PG *#u- reflecting the supporting vowel of PG *#uR- < PIE * -

In the following Proto-Germanic lexemes, PG *#u- should be regarded as the supporting vowel of
PG *#uR- < PIE *#(H) -, i.e. PG *#ul- < PIE * H l-, PG *#um- < PIE * H -, *#un- < PIE
* H n- and PG *#ur- < PIE * H r-.
83 PG *um(bi) adv./prefix/prep. ‘around about’. From PIE *h2 bhí, i.e. a fossilised instr.sg.
to the root noun PIE *h2ent- ‘face front’.
84 PG *un- prefix ‘un-’. From PIE *n-.
85 PG *unda prep. ‘until’. From PIE *h2nt-ó.
86 PG *unda, *under- etc. adv./prefix ‘under’. From PIE *ndh-o-, *ndh-ero-.
87 PG *under- prep. ‘among’. From PIE *h1n-tero- derived from PIE *h1en(i) ‘in’.
88 PG * n tw n - f. ‘dawn last part of the night’. From PIE *nkwt-, i.e. a zero grade
formation of the root PIE *nekwt- ‘night’.
89 PG *unka(n)- m. ‘snake’. From PIE *h2ngwh-on- with PIE *-gwh-n- > PG -kk- (> *-k-) by
means of Kluge’s Law.
90 PG *unnana- s.v. ‘love grant’. From PIE *h3n-n-h2-, i.e. a zero grade of the root PIE
*h3neh2-(?) ‘enjoy’.
91 PG *unþi- ~ nþ - f. ‘wave’; see (69).
92 PG *uns- pron. ‘us our’. From PIE *ns-, i.e. the oblique stem of the pers.pron. 1.pl.
93 PG *unsti- f. ‘storm’. From PIE *h2n 1-sti-, i.e. a zero grade of the root PIE *h2enh1-
‘breathe’.
94 PG *urba- n. ‘inheritance; piece of cattle’. From PIE *h3rbh-o-, i.e. a uniquely Germanic
zero grade formation next to normal PIE *h3orbh-o- which is also attested in Germanic as
PG *arba- m./n. ‘working animal worker(?) servant’.

3.5. Material from other sources (analogical reshapings, onomatopoeias, lexical borrowings)

The following Proto-Germanic lexemes contain an initial PG *#i-, *#ai-, * ī-, *#u-, *#au- or *#eu-
that may stem from other sources than those already mentioned, e.g. analogical reshaping, folk
etymology, onomatopoeia or lexical borrowing.
95 PG *ainia- m. ‘juniper’; see (3).

159
96 PG *a þīn- f. ‘mother’ *a þ n- f. ‘id.’. From PIE/Pre-PG *ait-, i.e. probably a Lallwort, cf.
e.g. Basque aita ‘father’ of with a similar structure, albeit obviously of different pedigree.
97 PG *aiþma- m. ‘son-in-law’; see (6).
98 PG *augan- n. ‘eye’. The radical vowel has undoubtedly been influenced analogically
from PG *ausan- n. ‘ear’; the expected development would rather be PIE *h3ekw-on- > PG
*agwan- > *agan- (or *awan-?), cf. Seebold (1967: 126-127).
99 PG *a j - (or perhaps rather PG *awjo-) f. ‘wetland island’. Developed regularly from PG
*agwj - < PIE *h2ekw- 2-, in itself a derivative of PIE *h2ékw-eh2- ‘water’.
100 PG *idis-/*edis- f. ‘lady’; see (26).
101 PG *iz- pron. ‘you your (pl.)’. The oblique stem of the pers.pron. 2.pl.; probably
analogically reshaped from PG *uz- (< PIE * s- ) on the basis of also analogically
reshaped nom. PG *jīz ‘you (pl.)’.
102 PG *īsarna- ~ *īzarna- n. ‘iron’. Lexical borrowing from PCelt. *īsarno-.
103 PG * j ja na- w.v. II ‘sound’; see (68).
104 PG *uhsan- m. ‘ox’; see (63).
105 PG * ww n-, * ww l n- etc. f. ‘owl’. Onomatopoetic; also PG *ūfa n - m. ‘id.’.

4. Ordering of data and preliminary conclusion

The following tables 2-9 will summarise the assumed prehistory of every Proto-Germanic lexeme
discussed in section 3.1, i.e. the estimated likelihood of each of them to continue a PIE form with
*#Hi- or *#Hu-.

Table 2: PG lexemes possibly reflecting PIE *#h1i-

Likely/possible Uncertain Unlikely/impossible

(4) PG *a s ja na- w.v. II ‘rush’ (1) PG *aima- m. ‘smoke steam; (2) PG *aina- num. ‘one alone any’
smell’

(8) PG *i- pron. ‘he she it’ (4) PG *a skr ja na- w.v. II ‘roar (2) PG *ainahan- adj. ‘single’
rage’

(9) PG *idi- f. ‘work’ (4) PG *īskr ja na- w.v. II ‘be (2) PG *ainaka- adj. ‘only special’
furious from excitement or pain’

(9) PG *ida- n. ‘constant moving (6) PG *aiþma- m. ‘son-in-law’ (2) PG *ainakj n- f. ‘widow’

160
quivering’

(9) PG * d ja na- w.v. II ‘move (10) PG *iliþ- m., lj - f. ‘footsole’ (2) PG *ainak(a)la- adj. ‘standing
around restlessly’ alone’

(13) PG *īwa- m. ‘yew’ (10) PG *ilkan- m. ‘footsole’ (3) PG *ainia- m. ‘juniper’

(12) PG *īsa- n. ‘ice’ (5) PG *aiþa- m. ‘oath’

(5) PG *aid(i)a- n. ‘isthmus’

(7) PG *aiwa- m., a w - f., *aiwi- f.

‘law’

(8) PG *ī- e.g. in Goth. ei conj. ‘that’

ON í (gær) ‘yesterday’ í (dag)
‘today’

(11) PG *īl ana- w.v. I ‘rush hurry’

(11) PG *īl - f. ‘hurry haste’

Table 3: PG lexemes possibly reflecting PIE *#h2i-

Likely/possible Uncertain Unlikely/impossible

(14) PG *aida- m. ‘pyre’ (14) PG *ai(d)ma- m. ‘smoke steam; (15) PG *aigana- pp.v. ‘own
smell’ possess have’

(14) PG *aidiana- w.v. I ‘burn (tr.) (14) PG *aima-uzj n- f. ‘embers’ (18) PG *aikana- s.v. ‘make one’s
harden with fire’ own; assign allot’

(14) PG *ai(d)la- n.(?) ‘flame’ (14) PG *a d skr ja na- w.v. II (19) PG *aikiana- w.v. I ‘annoy
‘roar rage’ pester’

(14) PG *ai(d)liana- w.v. I ‘burn (18) PG *aihtr ja na- w.v. II ‘beg (20) PG *aikwernan- m. ‘squirrel’
(tr.) ignite’ pray’

(14) PG *ai(d)lida- m. ‘fire’ (22) PG *aira-, *airu- m. (21) PG *aina- num. ‘one alone
‘messenger’ any’

(14) PG *a d s n- f. ‘forge (22) PG *a r n ja na- w.v. II ‘be a (21) PG *ainahan- adj. ‘single’
fireplace’ messenger, negotiate’

(15) PG *aigena- ~ *aigana- adj. (23) PG *aiskaþla- n. ‘heart’ (21) PG *ainaka- adj. ‘only special’
‘own’

(15) PG *aihti- f. ‘belongings, (26) PG *aiþma- m. ‘son-in-law’ (21) PG *ainakj n- f. ‘widow’

possessions property’

161
(15) PG *a g n- f. ‘ownership, (21) PG *ainak(a)la- adj. ‘standing
property’ alone’

(15) PG *aigni- f. ‘land property’ (22) PG *airi adv. ‘early’

(16) PG *aigena-, *aiginþ-? n. (22) PG *airiz adv.comp. ‘before

‘shoot barb’ earlier’

(16) PG *aigla- m. ‘shoot’ (25) PG *aiþa- m. ‘oath’

(17) PG *aik- f. ‘oak’ (25) PG *aid(i)a- n. ‘isthmus’

(17) PG *a kīna- adj. ‘oaken’ (27) PG *aiwa- m., a w - f., *aiwi- f.

‘age eternity’

(19) PG *aikala- adj. ‘excited (by (27) PG *aiwin, *aiwan adv. ‘ever
fear)’ always’

(19) PG *aikena- ~ *aikana- adj./ptc. (27) PG *a wīn- f. ‘eternity’

‘wild furious’

(20) PG *īkwernan- m. ‘squirrel’ (27) PG *aiwan- m. ‘eternity’

(23) PG *a sk - f. ‘demand (28) PG *aiwa- m., a w - f., *aiwi- f.

investigation’ ‘law’

(23) PG *a sk n- f. ‘question, (30) PG *aiza- n. ‘copper ore brass’

search investigation’

(23) PG *a sk ja na- w.v. II (32) PG *idi- f. ‘work’

‘demand inquire ask; investigate,
examine’

(23) PG *a sk ng - f. ‘demand’ (32) PG *ida- n. ‘constant moving

quivering’

(24) PG *aita- m. ‘abscess ulcer’ (32) PG * d ja na- w.v. II ‘move

around restlessly’

(24) PG *a st n- f. ‘testicle’ (33) PG *idis-/*edis- f. ‘lady’

(24) PG *aitila- adj. ‘swollen’ (34) PG *īdala- adj. ‘void idle,

futile’

(24) PG *aitra- n. ‘poison pus’

(29) PG *aiwiana- w.v. I ‘despise’

(29) PG *aiwiska- adj. ‘shameful’

(29) PG *aiwiskia- n. ‘shame

162
disgrace’

(29) PG *a w sk - f. ‘dishonour,
disgrace, offence’

(29) PG *a w sk ja na- w.v. II

‘make ashamed treat shamefully’

(31) PG *a z - f. ‘peace clemency;

respect, benevolence’

(31) PG *aiziana- w.v. I,

*a z ja na- w.v. II ‘forgive; honour’

(31) PG *a st ja na- w.v. III

‘respect’

Table 4: PG lexemes possibly reflecting PIE *#h3i-

Likely/possible Uncertain Unlikely/impossible

(35) PG *a r - f. ‘oar’ (36) PG *aiþa- m. ‘oath’

(36) PG *aid(i)a- n. ‘isthmus’

Table 5: PG lexemes possibly reflecting PIE *#Hi-, i.e. undeterminable timbre of the laryngeal

Likely/possible Uncertain Unlikely/impossible

(38) PG *īsa- m./n. ‘ice’ (37) PG *aibra- adj. ‘harsh’

(37) PG *ībra- m. ‘zeal eagerness’

(38) PG *īsa- m./n. ‘ice’

Table 6: PG lexemes possibly reflecting PIE *#h1u-

Likely/possible Uncertain Unlikely/impossible

(41) PG *(haima- l j n- f. ‘sorrel’ (39) PG *auda- m./n. ‘riches, wealth; (40) PG *auja- n.? ‘luck fortune
fate destiny’ wealth’

(43) PG *ūdra- n. ‘udder’ (39) PG *auda- adj. ‘rich’ (40) PG *awidi- *awida-, *auþa-,
*auþu-(?) adj. ‘easy comfortable’

(45) PG *eusizan- adj.comp. ‘better’ (39) PG *audaga-, *audiga- adj. (40) PG *aw d , *a þ adv. ‘easily’
‘rich’

(45) PG *e s l - f. ‘ease’ (39) PG *audena- ~ audana- adj. (41) PG *aula(n)- m. ‘fool (tall)

163
‘granted’ lanky fellow’

(46) PG *uba prep. ‘under; above’ (44) PG *e fn n- f. ‘crowd’ (41) PG *eula(n)- m. ‘(stalk of)
angelica silvestris’

(46) PG *upp adv. ‘up upwards’ (42) PG *auma- adj. ‘poor

miserable’

(46) PG * ban adv. ‘from above’ (43) PG *eudra- n. ‘udder’

(46) PG *ufuman- adj.comp. ‘highest, (44) PG *eup adv. ‘up upwаrds’

upmost’

(46) PG *uber- adv./prep. ‘over’

(46) PG * ft adv. ‘often’

(46) PG *ubez- w -) f. ‘sth. tall;

eaves’

(46) PG *ubila- adj. ‘evil bad’

(46) PG *ūba- adj. ‘ill-natured,

malicious’

(47) PG *unhta- adj. ‘accustomed’

(48) PG * sl - f. ‘embers glowing

ashes’

(48) PG * sj n- ~ *(aima-) zj n- f.
‘embers’

(48) PG *ustr(j)a- adj. ‘very active’

Table 7: PG lexemes possibly reflecting PIE *#h2u-

Likely/possible Uncertain Unlikely/impossible

(51) PG *auja- n.? ‘luck fortune (49) PG *au- prefix ‘away’ (52) PG *aukana- s.v. ‘increase,
wealth’ augment’

(51) PG *awidi- *awida-, *auþa-, (50) PG *auda- m./n. ‘riches, wealth; (52) PG *aukan- m. ‘increase,
*auþu-(?) adj. ‘easy comfortable’ fate destiny’ addition’

(51) PG *aw d , *a þ adv. ‘easily’ (50) PG *auda- adj. ‘rich’ (52) PG *a k ja na- w.v. II
‘increase augment’

(52) PG *aukena- ~ *aukana- ptc. (50) PG *audaga-, *audiga- adj. (54) PG *aula(n)- m. ‘fool (tall)
‘increased augmented’ ‘rich’ lanky fellow’

164
(53) PG *auk(e) adv./conj. ‘also; and’ (50) PG *audena- ~ audana- adj. (54) PG *eula(n)- m. ‘(stalk of)
‘granted’ angelica silvestris’

(56) PG *ūra- n. ‘soil(?)’ (54) PG *(haima- l j n- f. ‘sorrel’

(56) PG *ūr - m. ‘aurochs’ (55) PG *auma- adj. ‘poor

miserable’

(57) PG *auzena- ~ *auzana- ptc. (56) PG *aura- m. ‘moisture water’

‘scooped poured’

(58) PG *austera- adv. ‘east’ (56) PG *aura- m. ‘(moist) earth

soil’

(58) PG *a str n- f. ‘Easter’ (57) PG *ausana- s.v. ‘scoop pour’

(59) PG *auþia- adj. ‘remote, empty, (57) PG *a s n- f. ‘bowl; ladle’

desert, desolate waste; destroyed’

(59) PG *a þ - f. ‘desert’ (60) PG *ausan-? n. ‘ear’

(59) PG *auþiana- w.v. I ‘destroy’ (61) PG *uba prep. ‘under; above’

(60) PG *auzan- n. ‘ear’ (61) PG *upp adv. ‘up upwards’

(61) PG * ban adv. ‘from above’

(61) PG *ufuman- adj.comp. ‘highest

upmost’

(61) PG *uber- adv./prep. ‘over’

(61) PG * ft adv. ‘often’

(61) PG *ubez- w -) f. ‘sth. tall;

eaves’

(61) PG *ubila- adj. ‘evil bad’

(61) PG *ūba- adj. ‘ill-natured,

malicious’

(62) PG *ufna- m. ‘oven’

(63) PG *uhsan- m. ‘ox’

Table 8: PG lexemes possibly reflecting PIE *#h3u-

Likely/possible Uncertain Unlikely/impossible

(65) PG *uba prep. ‘under; above’ (64) PG *auzan- (~ *ausan-?) n. ‘ear’

165
Table 9: PG lexemes possibly reflecting PIE *#Hu-, i.e. undeterminable timbre of the laryngeal

Likely/possible Uncertain Unlikely/impossible

(66) PG *audena- ~ audana- adj. (66) PG *auda- m./n. ‘riches, wealth; (67) PG *auma- adj. ‘poor
‘granted’ fate destiny’ miserable’

(66) PG *auda- adj. ‘rich’

(66) PG *audaga-, *audiga- adj.

‘rich’

(68) PG * j ja na- w.v. II ‘sound’

(69) PG *unþi-, nþ - f. ‘wave’

(70) PG *ūt adv. ‘out’

(70) PG *uz adv./prep. ‘out’

Based on tables 2-9 above, it seems safe to assume that:

1 PIE *#h1i- > PG *#i-. Both PG *#ai- and PG *#i- are listed in the table, but with PG *#i-
as one of the two options, PG *#i- must be the expected outcome since it, unlike PG *#ai-,
cannot be explained in any other way than by PIE #h1i-;
2 a development of PIE *#h2i- > PG *#ai- cannot be secured, but it can be stated with a great
amount of certainty that no examples of PIE *#h2i-> PG *#i- can be found;
3 PIE *#h3i- may be represented in only one example in Germanic for which reason the
statistical evidence does not allow for a statement as to the Germanic outcome;
4 PIE *#Hi- comes with too vague examples;
5 PIE *#h1u- > PG *#u-. Both PG *#eu- and PG *#u- are listed in the table, but with PG
*#u- as one of the two options, PG *#u- must be the expected outcome since it, unlike PG
*#eu-, cannot be explained in any other way than by PIE #h1u-;
6 a development of PIE *#h2u- > PG *#au- cannot be secured, but it can be stated with a
great amount of certainty that no examples of PIE *#h2u-> PG *#u- can be found;
7 PIE *#h3u- may be represented in only one example in Germanic for which reason the
statistical evidence does not allow for a statement as to the Germanic outcome;
8 PIE *#Hu- comes with too vague examples; and
9 PIE *#HiH- > PG * ī- and PIE *#HuH- > PG * ū- regardless of the timbre of the
laryngeals as examplified by (13) PG *īwa- m. ‘yew’ (20) PG *īkwernan- m. ‘squirrel’

166
(38) PG *īsa- m./n. ‘ice’ and (43) PG *ūdra- n. ‘udder’ (56) PG *ūra- n. ‘soil(?)’ (56) PG
*ūr - m. ‘aurochs’ respectively.

It might be worth attaching a comment or two on the proposed development of PIE *#h2i- > PG
*#ai- and of PG *#h2u- > PG *#au-. As I stated above, it is true that this development cannot be
established with absolute certainty. Various analogical and other processes may simply have blurred
the picture considerably. However, I personally find it remarkable that, out of 33 possible or likely
examples of PIE *#h2i-, 32 contain PG *#ai-, the remaining one containing PG *#ī- < PIE *h2iH-.
Not a single example contains PG *#i-. Correspondingly, 12 out of the 14 examples for which PIE
*#h2u- has been judged possible or likely contain PG *#au-, the remaining two containing PG *#ū-
< PIE *h2uH-. Again, we find no examples with a short monophthong, i.e. with PG *#u-. Needless
to say, many of the examples may contain regular or analogically arisen full or o-grade, but I would
be utterly surprised and find it statistically significant if not even a single of these examples would
have descended from PIE *#h2i- or *#h2u- by means of regular sound change.
If we can thus conclude that the suggested sound changes of PIE *#h2i- > PG *#ai- and PIE
*#h2u- > PG *#au- are indeed quite likely, a new question almost automatically arises, viz. why a
parallel development cannot be posited for PIE *h1 and *h3. Honestly, we cannot estimate with
certainty the development of PIE *#h3i- and PIE *#h3u-. Consequently, we cannot exclude the
possibility of these sequences also yielding forms with initial diphthongs in Germanic. As for PIE
*h1, however, there is no doubt that the Germanic outcome was an initial monophthong, i.e. PG *#i-
and *#u-. In my opinion, we may find the reason for this discrepancy between the developments of
at least PIE *h1 and *h2 in the circumstance already mentioned (cf. section 3) that the forms with
initial diphthong might have arisen in a specific sandhi environment, viz. the sequence PIE
*-C#Hi/uC- > *-CHi/uC- where a supporting vowel could be developed in order to ease the
pronunciation, i.e. PIE *- HH C-. From that point of view, it is easy to understand why a PIE *h2
with a pronunciation probably on the lines of [x], cf. e.g. Rasmussen (1983 [1999]: 77), would be
considerably more prone to generating a supporting vowel i.e. [Cəxi/u] vel sim. than PIE *h1 with
a pronunciation probably on the lines of [h] which would probably just be eliminated when adjacent
to a consonant, i.e. [C(h)i/u] vel sim.

167
5. Excursus: PG *ubila- ‘evil, bad’ and *ufna- ‘oven’ – why not †aubila- and †aufna-?
One of the preliminary conclusions presented in the previous paragraph, viz. that PIE *#h2u- did not
develop into PG *#u- and cannot be said not to have developed into PG *#au-, relies on the
premises that PG *ubila- adj. ‘evil bad’, including maybe also PG *uba prep. ‘under; above’ and
PG *ufna- m. ‘oven’ have not developed from a form with a word-initial *h2. Thus, I have
suggested that PG *ubila- adj. ‘evil bad’ and PG *uba prep. ‘under; above’ < PIE *h1up- and that
PG *ufna- m. ‘oven’ is in fact a Wanderwort together with its pseudo-cognates of Hitt. huppar n.
‘bowl’ Gr. ἱπνός, ἰπνός m. ‘oven’ Myc. i-po-no- m. ‘dutch oven i.e. earthenware bowl used for
baking on a hearth’ and OPr. wumpnis sb. ‘baking oven’.
Not every scholar would accept these premises. PG *ubila- adj. ‘evil bad’ including maybe PG
*uba prep. ‘under; above’ is often seen reconstructed as PIE *h2up- and thus compared to Hitt.
huwappa- adj. ‘evil ill bad’ derived from huwapp- ~ hupp- v. ‘be hostile towards, do evil against;
throw (down) hurl’. Similarly PG *ufna- m. ‘oven’ is often reconstructed with initial PIE *h2, i.e.
PIE *h2up-no-, in the light of its obvious, semantic connection with Hitt. huppar n. ‘bowl’ and the
formal difficulties concerning the comparison of these two forms to Gr. ἱπνός, ἰπνός, Myc. i-po-no-
and OPr. wumpnis are, if not disregarded, then at least heavily downplayed. If, as indicated by the
list of forms in table 7 above, PG *#au is actually the regular result of PIE *#h2u-, we would have
expected PIE *h2upiló- to yield PG †aubila adj. ‘evil bad’ rather than *ubila- and PIE *h2úpno- to
yield PG †aufna- m. ‘oven’ rather than *ufna-. Consequently, the only possible solution for scholars
not accepting the etymologies proposed by me would seem to be that PIE *#h2u- > PG *#u-, i.e.
what has been presented here as the communis opinio, in spite of all the fitting candidates for PIE
*#h2u- > PG *#au-.
The regular sound change presented by Hyllested & Cohen (2007: 13) for Greek, viz. that there
are “[…] no examples in Greek of u-diphthong + a labial reflecting either PIE full-grade *HewP- or
PIE zero grade *HuP- in initial position (where P = any labial, i.e. any of /p, b, b h m/).” may
actually serve as inspiration for a compromise between those advocating for PG *ubila- < PIE
*h2upiló- and PG *ufna- < PIE *h2úpno- and those believing in the possibility of a default
development of PIE *#h2u- > PG *#au-.
Even a mere browse through the entire Proto-Germanic corpus will reveal that Proto-Germanic
offers conditions comparable to those of Greek. In Proto-Germanic, it turns out, we find almost no
examples of a u-diphthong followed by a labial consonant in initial position. In fact, no more than
three counterexamples can be found.

168
1 PG *auma- adj. ‘poor miserable’ with derivatives. Only attested in North Germanic and
probably to be regarded as a spontaneous dissimilation from PG *arma- adj. ‘id.’ <
*arbma-. See (67) for additional details.
2 PG *e fn n- f. ‘crowd’. Probably derived from PG *eup adv. ‘up upwards’ cf. below;
alternatively seen as etymologically enigmatic. See (44) for additional details.
3 PG *eup adv. ‘up upwards’. Only attested in Gothic and possibly secondarily backformed
from the locative PG *uppai adv. ‘up upon above’ cf. Kroonen (2010: 374-376; 2013:
121). See (44) for additional details.

Whereas PG *auma- adj. ‘poor miserable’ can thus easily be dismissed PG *eup adv. ‘up
upwards’ constitutes a considerably stronger counterexample. Despite Kroonen’s (2010: 374-376;
2013: 121) attempt to explain it as secondary from PG *uppai adv. ‘up upon above’ I fail to see
any phonological, morphological or other motivation for the introduction of full grade vocalism and
therefore must suspect that the full grade represented in PG *eup is original and archaic, cf. also
Kroonen’s (2010: 375-376) own reference to the partly similar situation found in directional
adverbs in Hittite where locative adverbs with radical zero grade, e.g. Hitt. ar adv. ‘forwards’ (<
PIE * r-ó), are occasionaly matched by allatives with radical full grade, e.g. Hitt. ran adv.
‘before’ (< PIE * r- ).
With PG *eup thus being, in fact, an example of a u-diphthong followed by a labial consonant,
we cannot apply Hyllested & Cohen’s (2007: 13) constraint for Greek on Proto-Germanic without
any amendments unless PG *eup is really secondary. For Proto-Germanic, the constraint would
have to be limited to the u-diphthong with PG *a as its vocalic element, i.e. PG *au > PG *u /
#_C[+lab] represented by PG /f, p, b, m/. Consequently, it seems safe to assume that a PG *aubila
adj. ‘evil bad’ (< PIE *h2upiló-) and a PG *aufna- m. ‘oven’ (< PIE *h2úpno-) would automatically
yield PG *ubila- and PG *ufna-, respectively, i.e. the Proto-Germanic reconstructed forms
underlying the forms actually attested in the ancient Germanic languages.
As a closing matter of curiosity, it also deserves mentioning that this or a similar constraint was
reintroduced in English in connection with the Great Vowel Shift as exemplified by, e.g., OE rū n.
‘room’ > ME roum > Eng. room /rūm/ not Eng. †/rawm/ as otherwise expected, cf. Hyllested &
Cohen (2007: 13-14).

169
6. Conclusion
In this article, I have demonstrated that, contrary to common belief, we can neither state with
certainty nor suggest tentatively that PIE *#Hu- > PG *#u-. Neither can we state nor suggest that
PIE *#Hi- > PG *#i-. Based on the analyses presented in this article, it would rather seem that:

1 PIE *#h1i- > PG *#i-. Both PG *#ai- and PG *#i- are listed in the table, but with PG *#i-
as one of the two options, PG *#i- must be the expected outcome since it, unlike PG *#ai-,
cannot be explained in any other way than by PIE #h1i-;
2 a development of PIE *#h2i- > PG *#ai- cannot be secured, but it can be stated with a great
amount of certainty that no examples of PIE *#h2i-> PG *#i- can be found;
3 PIE *#h3i- may be represented in only one example in Germanic for which reason the
statistical evidence does not allow for a statement as to the Germanic outcome;
4 PIE *#Hi- comes with too vague examples;
5 PIE *#h1u- > PG *#u-. Both PG *#eu- and PG *#u- are listed in the table, but with PG
*#u- as one of the two options, PG *#u- must be the expected outcome since it, unlike PG
*#eu-, cannot be explained in any other way than by PIE #h1u-;
6 a development of PIE *#h2u- > PG *#au- cannot be secured, but it can be stated with a
great amount of certainty that no examples of PIE *#h2u-> PG *#u- can be found;
7 PIE *#h3u- may be represented in only one example in Germanic for which reason the
statistical evidence does not allow for a statement as to the Germanic outcome;
8 PIE *#Hu- comes with too vague examples; and
9 PIE *#HiH- > PG * ī- and PIE *#HuH- > PG * ū- regardless of the timbre of the
laryngeals.

Basing my estimation on statistical evidence, I would even dare stating that it would be statistically
significant if not even a single of the many examples of PG *#ai- and *#au- deemed possible or
likely of continuing PIE *#h2i- and *#h2u- does not also continue PIE *#h2i- and *#h2u- in reality.
Furthermore, if we may draw partial parallels to Germanic from the developments seen in Greek
and English, viz. that any example of PG *#au- followed by a labial consonant would result in PG
*#u-, it would seem that the conclusions presented above hold good regardless of the pedigree of
PG *ubila- adj. ‘evil bad’ *uba prep. ‘under; above’ and *ufna- m. ‘oven’.

170
Article no. 4: The outcome of PIE *- and *-
in Germanic

By suggesting an interconnected series of soundlaws for the outcome of Proto-Indo-

European falling e-vowel diphthongs in final syllables in Proto-Germanic and in the
individual Germanic languages, viz. PIE *-e > PG *-ai(C)#, PIE *- > PG
*-ei(C)#, PIE *-e > PG *-au(C)#, and PIE *- > PG *-eu(C)#, this
44
article renders superfluous the old, prevalent assumption of competing o-grade
allomorphs in some of the oblique cases of the PIE i- and u-stems. Consequently, the
i-stem gen.sg. is reconstructed only as PIE *-e s (not as †-ois in addition), the u-stem
gen.sg. only as *-e s (not as †-o s), the u-stem loc.sg. only as *- (not as †- ), the
u-stem voc.sg. only as *-e (not as †-o ), etc.

1. The importance of the Germanic “Auslautgesetze”

To many scholars of Indo-European linguistics, the question of the Germanic “Auslautgesetze”
constitutes a frustrating puzzle and an endless sequence of circular argumentation that may actually
raise more questions than it answers. Meillet (1922: 79-80) stated these concerns quite succinctly in
his Caractères généraux des langues germaniques: ”Le traitement des finales est imparfaitment
connu. Les exemples de chaque type de faits sont rares et l’original indo-europ en n’est pas
toujours déterminable. On opère avec des formes grammaticales, suspectes par leur nature même
d’avoir subi des actions analogiques.”
Even if the situation has indeed improved since Meillet’s days many issues still lack tenable
solutions; solutions that the scholarly community definitely ought to seek, for when carefully
examining the different Proto-Germanic outcomes of Proto-Indo-European final syllables, i.e. the
syllables containing the desinences, we soon come to realise that a correct and thorough
understanding of the Germanic “Auslautgesetze” may actually contribute considerably to the
solutions to long-lasting problems of PIE phonology and desinential morphology. One such case is
the question of the development of word-final diphthongs where the scholarly community has, so
far, posited a range of adversely and unmotivatedly competing PIE desinences.

44 A slightly modified version of this article has been accepted for publication in the peer-reviewed journal NOWELE,
vol. 67/2 (2014), pp. 149-172.

171
The present study offers a new explanation of the development of the PIE -vowel diphthongs in
final syllables. The corresponding ā/ō-vowel diphthongs will play only a minor role. What is, in
other words, the expected outcomes of PIE *-e (C)#, *- , *-e , and *- ? When in
final position, precisely these diphthongs play a pivotal role in our understanding of the desinential
morphology of the PIE i- and u-stems in particular. In fact, they are found nowhere else in the PIE
desinential system except for in the mainly Greek (and Baltic) α - or ἱππεύ -type and its
seeming parallel type πάτρως (< *- s).45

2. The Proto-Indo-European state of affairs

The oblique cases of the PIE i- and u-stems may be formed in one of two ways: either by the
derivational suffix in the unaccented zero grade represented by the glide or (corresponding to the
i or u of the stem) followed by the standard desinence in the accented full grade, e.g. i-stem gen.sg.
*- -é/ós or u-stem dat.sg. *- - (open inflection) or by accented ‘full grade’ of the derivational
suffix followed by unaccented zero grade of the desinence proper, e.g. i-stem gen.sg. *- -s or u-
stem dat.sg. *- -e (closed inflection). Only the latter principle, which was also by far the one
most frequently adopted in the formation of oblique cases of PIE i- and u-stems, is of relevance
here.
The existence of these two types was known and recognised fairly early, cf. e.g. Wackernagel &
Debrunner (1929: 138-144) who also very briefly mentions a third type represented, in Indic, only
by Skt. sá k - ‘friend companion’. To my knowledge however serious attempts at an
explanation of the existence of the two types were not provided until Szemerényi (1970: 160-165)
suggested that the subsidiary type (with gen.sg. in *- -os, *- -os or type I in Szemer nyi’s terms)
would originally have contained only nouns whose underlying stem contained an open syllable,
whereas stems in underlyingly closed syllables would originally have yielded the prevalent type
(with gen.sg. in *-e -s/-o -s, *-o -s/-e -s or type II in Szemer nyi’s terms). Stems of the structure
CeC-i/u- would thus yield oblique cases of type I whereas stems of the structure CeCC-i/u- or CeC-
Ci/u- would result in oblique cases of type II.46

45 It has been suggested that this type did not, in fact, contain an original diphthong, cf. e.g. Szemerényi (1957: 159-
181), Schindler (1976: 351-352), Rasmussen (1989: 273-274) and Olsen (2008).
46 This distributional theory has the additional advantage of providing a rationale for the preference of type II to type I.
Most i- and u-stems are not primary but, in fact, secondary, e.g. ti- and tu-stems. Such secondary stems would

172
Table 1: Sze er n ’s 1970: 162 or g nal reconstr cted des nences of t e II:

i-stems u-stems

nom.sg. *-i-s *-u-s

acc.sg. *-i-m *-u-m

gen.sg. *-ei-s / *-oi-s *-ou-s / *-eu-s

dat.sg. *-ei-ei *-ou-ei / *-eu-ei

loc.sg. *- i47 *-ōu / *- u

voc.sg. *-ei *-ou

nom./voc.pl. *-ei-es *-ou-es / *-eu-es

If we were to believe the linguistic data represented in table 1, it would seem incontrovertible that,
in a number of cases, the PIE speakers could choose between two competing desinences, e.g. i-stem
gen.sg. *-e -s vs. *-o -s and u-stem nom./voc.pl. *-o -es vs. *-e -es. Needless to say, unregulated
allomorphy – as well as free variation in general – is far from being an unknown typological
phenomenon in the world of linguistics,48 but we should always try to provide realistic explanations
for the variation, e.g. if any variant is more original. Szemerényi (1970: 163-165) actually did this.
He proposed an original distribution with e-vowel pertaining to the oblique cases of the i-stems and
o-vowel pertaining to the oblique cases of the u-stems. He further believed this *e and *o to have
developed from earlier **i and **u, respectively, i.e. PIE **- - > *-e - and **- - > *-o -, and he
regarded the i- and u-vowels as resulting from some kind of pre-PIE Sievers development where an

virtually automatically produce closed stem syllables seeing that the general, minimal root structure of PIE is CeC,
cf. e.g. Benveniste (1935: 170).
47 Szemerényi (1970: 110) explains the long grade loc.sg. form PIE *- as the result of earlier **-e -i. The
corresponding u-stem form PIE *- and, secondarily according to Szemerényi, *- would then have emerged as
the result of proportional analogy to the i-stem desinence. The original u-stem loc.sg., i.e. the form not originating
from analogical processes, might, then, be what underlies the Skt. by-form -avi < PIE *-e -i, cf. e.g. Wackernagel &
Debrunner (1929: 157) and Kuiper (1942: 214-215).
48 We need not go any further than to other parts of the PIE desinential system where, only to mention one example,
free variation seems to have prevailed between *m and *bh in the dat./abl.pl., instr.pl., and dat./instr./abl.du. of any
PIE nominal paradigm but the thematic one (with instr.pl. in PIE *- s rather than †-o-mis/-o-bhis), cf. e.g. Fortson
(2004: 116). Later, of course, the former variant was generalised in Germanic and Balto-Slavic, the latter variant in
the remaining branches.

173
i or a u would be inserted before a homorganic glide in order to facilitate the pronunciation of heavy
consonant clusters. In this way, a straightforward basis for mutual analogical levelling between the
morphologically parallel i- and u-stems has been created so as for u-stems to add the e-vowel forms
(*-e -) to the existing inventory of o-vowel forms (*-o -) and vice versa for the i-stems. Elegant
and ingenious as this explanation may seem, it fails to account for the precise nature of this
otherwise unattested pre-PIE Sievers development as well as for the fact that e-vowel forms have
infected the u-stem paradigm to a significantly higher extent than o-vowel forms have infected the
i-stem paradigm. Thus, as can be seen from table 1, Szemerényi lists four secondary e-vowel forms
in the u-stem paradigm (gen.sg. *-e -s, dat.sg. *-e -e , loc.sg. *- , and nom./voc.pl. *-e -es) but
only one secondary o-vowel form in the i-stem paradigm (gen.sg. *-o -s). In my view, this
distribution can suggest hardly anything but a general preference for e-vowel forms to o-vowel
forms.
Rasmussen (1996: 137-141) attempts to account for at least the first of the difficulties of
Szemer nyi’s model viz. the precise nature of this otherwise unattested pre-PIE Sievers
development. By positing that, underlyingly, the derivational suffixes of the i- and u-stems were not
the ‘naked’ vowels *-i- and *-u- alone but rather the corresponding glides *- - and *- - preceded by
the suffixal full grade e-vowel, i.e. *-e - and *-e -, Rasmussen renders the assumption of the said
pre-PIE Sievers development superfluous. With application of a process where all unstressed
vowels are deleted at a pre-PIE stage, the stem would then initially have appeared in the form of
* - - or * - - to which the true case forms should be added. However, Rasmussen
argues that this pre-PIE stage would not allow stems ending in three or more consonants;49
consequently, a structure like * - - would not be allowed, and the underlying e-vowel of the
derivational suffix would be retained, i.e. * -e - rather than the structurally inadequate
† - -. When regularly shifting the accent one slot rightwards in the oblique cases of the two
different types of i- and u-stems, i.e. * - - with expected deletion of the suffixal vowel and
* -e - with retention of the suffixal vowel in order to avoid the stem ending in three
consonants, we would end up with e.g. gen.sg.-forms in * e - -é/ós and *CeCC- -s or
*CeC-C -s, respectively, from which the forms normally reconstructed for PIE are easily

49 Rasmussen (1996: 581) supports his argument by calling the readers’ attention to the fact that no PIE suffix /-nt-/
exists, only /-ent-/ added to a preceding stem final or root final consonant.

174
deducible by application of yet another round of deletion of unstressed syllables and subsequent,
paradigmatic levelling of the zero grade of the root from the oblique cases into the strong cases.
However given Rasmussen’s model of positing an underlying e-vowel in both the i- and u-
stems, no account can be provided for the presence of o-vowel forms. Admittedly, such a suffixal o-
grade could have arisen in the voc.sg. or nom./voc.pl., which, in a proterokinetic paradigm such as
the one posited for the type II i- and u-stems, would be accented on the root syllable with
weakening or at least alteration of the suffixal vowel as a consequence but Rasmussen’s article
does not mention that option. It should also be noted that, even if the type II i- and u-stems are
basically proterokinetic, they fail to show the expected, unaccented o-vocalism of “true”
proterokinetic paradigms in the two most prominent cases, viz. nom.sg. and acc.sg. Such o-grade
forms are found merely within the residual sak -type mentioned above, cf. e.g. Kuiper (1942:
197-198).50 For the sake of convenience, we shall therefore regard all case forms of the type II i-
and u-stems except for the nom.sg. and acc.sg. as oblique or weak even if this is not in accordance
with standard terminology.

50 As for the system of Proto-Indo-European accent and ablaut, I choose to follow Rasmussen (1978 [1999]: esp. 17-
20, 27-28, 44-45 LG 1+3+7) who operates with and elaborates on the theoretical model of the Erlanger Schule. He
thus claims that e-grades are accented and that unaccented original PIE *e are first turned into *o only to be lost
altogether at a later stage (zero grade). However, if lengthened by means of influence from the *-s of the PIE
nom.sg. or the *-h2 of the PIE collective, the o would be preserved. Hence follows that an original proterokinetic
paradigm would be nom.sg. *CéC-eC-s > *CéC-oC-s > *CéC- -s, acc.sg. *CéC-eC-m > *CéC-oC-m > *CéC-C-m
( *CéC-oC-m in analogy with the pattern of the hysterokinetic paradigm type) and gen.sg. *CeC-éC-e/os > *CoC-
éC-e/os > *CC-éC-s, whereas an original hysterokinetic paradigm would be nom.sg. *CeC-éC-s > *CoC-éC-s >
* o - -s > * - -s, acc.sg. *CeC-éC-m > *CoC-éC-m > *CC-éC-m and gen.sg. *CeC-eC-é/ós > *CoC-oC-é/ós
> *CC-C-é/ós. From this follows that the residual i-stem type of Skt. sak - is in principle a ‘true’ proterokinetic
i-stem with nom.sg. in PIE*- s. The two prevalent types of i- and u-stems, however, seem to resist analysis within
the framework of this theoretical model since we would a priori expect the type I (open inflection with gen.sg. in
PIE *- -ós/- -ós) to have a nom.sg. †- -s/†- -s and an acc.sg. †- -m̥/†- -m̥, not *-is/-us and *-im/-um as actually
attested, and the type II (closed inflection with gen.sg. in PIE *- -s/*- -s) to have a nom.sg. †- -s/†- -s and an
acc.sg. †- -m̥/†- -m̥, not *-is/-us and *-im/-um as actually attested. Rasmussen (1978 [1999]: 42) seeks to explain
this discrepancy by means of prevocalic sandhi variants where original (i.e. before the series of soundlaws
mentioned above came into effect) nom.sg. PIE *-e s/-e s - was reinterpreted as *-e /-e s - (> *-o /-o s - > *-i/-
u sV- by application of the same soundlaws). For a different account of the prehistory of the i- and u-stems cf. e.g.
Beekes (1985: esp. 150-164).

175
To my knowledge, the scholarly community has yet to propose a theoretical model that can
account smoothly for the suffixal o-grade forms of the oblique cases of the PIE i- and u-stems.51
Consequently, we should ask if there were, in fact, any o-grade forms at all. At first sight, any such
speculation would seem futile seeing that, after all, suffixal o-grade forms of the oblique cases of
the i- and u-stems seem to be attested in most Indo-European branches, viz. in Indo-Iranian, Balto-
Slavic, Italic, Celtic, Anatolian and Germanic.

3. Apparent suffixal o-grade forms in the individual branches

At first glance, Indo-Iranian might be the easiest branch to account for. As a consequence of the
Indo-Iranian merger of PIE * , * , and * into * , cf. e.g. Wackernagel & Debrunner (1896: 4-5), it
is simply not possible to determine if a given Proto-Indo-Iranian diphthong * would have resulted
from PIE * , * or * and, correspondingly, if a given PIIr diphthong * would have resulted
from PIE * , * or * . Only in one position does a difference arise between the outcomes of
(short) PIE *e, *a, and *o in PIIr,52 viz. in open syllables where Brugmann’s Law dictates that PIE
*o > PIIr * rather than *a, cf. e.g. Wackernagel & Debrunner (1896: 13-14). For the i- and u-
stems, such surroundings are found only in the dat.sg. (PIE *-e -e /*-o -e ) and nom./voc.pl. (PIE
*-e -es/*-o -es) where the Indo-Iranian ubiquity of the short-vowel diphthong *-au- and the
consequent absence of the corresponding long-vowel diphthong *- - strikingly point towards PIE
*-e - rather than *-o -, e.g. Skt. sūnáve ‘son’ (dat.sg.).53 The PIIr i-stem loc.sg. *- 54
(and u-stem

51 Beekes (1985) actually operates with only e-grade forms in the oblique cases of the PIE i- and u-stems. For instance,
he, as one of the few scholars dealing with this problem, reconstructs only e-grade forms in the gen.sg. (Beekes
1985: 128). However, to my belief, Beekes still fails to account for the appearance of the alleged o-grade forms.
Bammesberger (1990: 126) reconstructs PG *-aiz as PIE *-o s but doubts that this form could really be of Proto-
Indo-European age.
52 Though irrelevant to the i- and u-stems due to lack of decisive material, a difference of outcome is also found where
palatalisation of a preceding velar plosive would reveal the origin of a PIIr *a as PIE *e rather than as *a or *o, cf.
e.g. Wackernagel & Debrunner (1896: 139-144).
53 An instance of PIIr *- - may be claimed for the Iranian by-form nom.pl. Av. - , OPers. - va, cf. e.g. Hoffmann
& Forssman (1996: 131). However, this form may be seen as analogical from the residual proterokinetic type of Skt.
sak - where o-grade vowels are, indeed, expected. Thus, for the sak -type, which is also attested in Iranian, we
would expect PIE nom.sg. *- s, acc.sg. *-o etc., cf. Kuiper (1942: 197-198) and also Beekes (1985: 85-89) for
parallel assumptions regarding some of the Iranian singular forms with .

176
loc.sg. *- ) can be explained from PIE *- and *- (and *- and *- ) alike, and the PIIr i-stem
gen.sg. *-ais (and u-stem gen.sg. *-aus) and i-stem voc.sg. *-ai (and u-stem voc.sg. *-au) that all
contain tautosyllabic glides can likewise be developed from PIE e- and o-vowel forms alike.
Consequently the remaining Indo-European branches must be the ones to show whether suffixal -
or ō-grade forms should be reconstructed for the i- and u-stem gen.sg., loc.sg. and voc.sg.
Balto-Slavic seems to offer suffixal o-grade forms in some of the oblique cases of the u-stems,
e.g. Lith. gen.sg. -aũs, voc.sg. -aũ and nom.pl. -aus/-ous (dialectal) for the Baltic material, and OCS
gen.sg. -u, dat.sg. -ovi, loc.sg. -u, voc.sg. -u, gen./loc.du. -ovu, nom.pl. -ove and gen.pl. -ovъ for the
Slavic material. Stang (1966: 73-75, 215-216), however, is among a number of scholars who are not
convinced that these outcomes actually mirror original PIE *-o -. If, as assumed for Balto-Slavic by
e.g. Kortlandt (1979: 57; 2008: 8) who elaborates on Stang’s theories (1966: 73-75, 215-216) for
Baltic, Proto-Balto-Slavic *e is rounded to *o before intervocalic *w, i.e. PIE *-e - > PBS *-ewV-
> *-owV-, we could easily eliminate most instances of apparent, suffixal o-grade vowels in the
oblique cases of u-stems in Balto-Slavic. The u-stem gen.sg. PBS *-aus, loc.sg. *-au/- ,55 and
voc.sg. *-au would have to be explained by paradigmatic levelling from the already affected
cases,56 as actually recognised already by Vaillant (1958: 110). With his claim that PIE * > PBS
* before taotsyllabic * , Olander (2014: 65-66, 147-148, 169-170, 196-197, 254-256) also fails to
see the need for the reconstruction of o-grade variants of the oblique cases of the i- and u-stems.

54 In Sanskrit, the original PIIr i-stem loc.sg. *- was replaced by the corresponding u-stem desinence, but it still
prevails in Iranian as Av. - in gәrәzd ‘in/by walking’ and maybe also as a sandhi variant - in Sanskrit, cf. e.g.
Hoffmann & Forssman (1996: 134).
55 An alternative u-stem loc.sg. -o̥u/-ū is found in Žemaitic and High Latvian. Stang posits a development -o̥u/-ū
[Stang’s notation] < *-uo < PIE*- which, if correct, implies the necessity of positing an original o-coloured vowel
in the u-stem loc.sg. Stang (1966: 215-216) recognises himself, though, that -o̥u/-ū might have arisen from the
enlarged form - oj .
56 If we were to accept the theories of handbooks such as Vaillant (1950: 110) who argues that the soundlaw PIE
*-e - > PBS *-ewV- > *-owV- would only happen if V = V[-front], we would also need to count the dat.sg. and the
nom.pl. to the cases with paradigmatic levelling in favour of PBS *-au-. In a recent article, Olander (2012) posits a
special development of original PIE *-os > Proto-Slavic (PS) *-әs, even when a glide is inserted in between, i.e. PIE
*-o s > pre-PS *-a x > PS *-ә > e.g. OCS -i, cf. especially Olander (2012: 332-333). Olander has further suggested
(p.c.) that a parallel development would have taken place with a u-diphthong, i.e. PIE *-o s > pre-PS *-a x > PS *-
ә > e.g. OCS -y. Since the OCS u-stem dat.sg. is, in fact, -u rather than †-y (< *-ū), it is seen that the origin of that
desinence cannot be PIE *-o s. More fruitful speculations would result in a development along the lines of PIE *-
e s > PS *-a x (postdating the Slavic development of *a > *ә before final *s) > PS *-a > e.g. OCS -u.

177
Similar soundlaws may be posited with certainty for Italic and with probability for Celtic. It is a
well-established fact that PIE *e yielded PIt. *ou in every position of the word (as in PIE *ne os
‘new’ > Lat. novus), cf. e.g. Sihler (1995: 40) for Latin and Buck (1904: 46) for Sabellic. As such,
accounting for u-stem case forms such as OLat. u-stem dat.sg. -uei (> Lat. - ī), Osc. gen.sg. -ous,
and Umbr. dat.sg. -u/-o (in trifu/trifo) as developed from forms with original e-vowel diphthongs is
entirely straightforward. The Celtic situation is just as propitious. Thurneysen (1975: 39-40, 122)
operates with a soundlaw PIE *e > OIr. ou (> ó in stressed syllables in Old Irish), by which
process OIr. u-stem gen.sg. -o (Ogham -OS) may be derived from PIE *-e -s as well as from PIE
*-o -s, and Lambert (2003: 44), among others, apply an identical soundlaw to the Gaulish material,
i.e. PIE *e > Gaul. ou. Consequently, the apparent o-grades of the Gaul. u-stem dat.sg. -οου and
nom.pl. -oves, cf. Lambert (2003: 62) and also Pedersen (1913: 91), may be ascribed to the same
phenomenon as the corresponding desinences of Old Irish and Italic. The Old Irish i-stem gen.sg. -
o/-a (Ogham -OS) deserves mentioning here, too, since it is, by some scholars, e.g. Pedersen (1913:
94), reconstructed as PIE *-o s, albeit with the questionable assumption that word-final PIE *-o s is
treated differently than PIE *-o , cf. the OIr. o-stem nom.pl. -i < PIE *-o . Alternatively, we could
ask if PIE *-e s could not simply have been remodelled as *-o s in analogy with the u-stems whose
*-o s has developed from PIE *-e s by regular soundlaw. After all, a high level of parallelism
prevails between these two stem types in general, cf. e.g. Pedersen (1913: 94) – who even advocates
himself that OIr. -o/-a < PIE *-o s by means of regular sound change – on the interparadigmatic
levelling between the gen.sg., the gen.du. and the gen.pl. forms of the stem types in question.
The o-grades of Anatolian cannot be as easily dismissed as those of the previous branches. To
my knowledge, no soundlaws can explain e.g. the Hitt. i-stem gen.sg. -a a or the u-stem gen.sg.
-awa from PIE *-e s and *-e s, respectively (with subsequent addition of the productive gen.sg.
-a < PIE *-os).57 According to e.g. Kimball (1999: 213-214, 220-221), original short-vowel
diphthongs are generally monophthongised: PIE *-e > Hitt. -e or -i (no consensus), PIE *-o > Hitt.
-e and PIE *-e /*-o > Hitt. -u. A sequence Hitt. -ai- could only result from an original long-vowel
diphthong, and – again according to Kimball (1999: 226-230) – probably only from the o-grade
diphthongs PIE *- and *- , since PIE *- > Hitt. - and PIE *- > Hitt. -ū.58 Ironically, Hitt.

57 In general, suffixal a-vowels are prevalent throughout the entire i- and u-stem paradigm, though competing with the
zero grades -(i)y-/-(u)w-.
58 Some scholars do, however, propose a change of PIE * > Hitt. au, cf. Kimball (1999: 230).

178
*-a < PIE *- s and Hitt. *-a < PIE *- s might be exactly what we need in order to establish
that Anatolian provides no evidence for suffixal o-grades of the oblique cases of the i- and u-stems.
Weitenberg (1984: 352-356, 369-376) mentions that Hittite tends towards generalising the full
grade suffix - - of the u-stem nom.sg. form throughout the entire paradigm, and we have to bear in
mind that this is the residual sak -type (represented in Hittite by e.g. l nga ‘oath’ or the neuter
plural-collective hastai ‘bone(s)’) – or, more precisely, the parallel type with nom.sg. in PIE *- s
(represented in Hittite by e.g. arna ‘birthstool’), cf. also e.g. Kronasser (1962-1966: 202-208,
250-251). These two types are far more predominant in Anatolian than in any other branch of Indo-
European. Granted the validity of that assumption, the Anatolian o-grade forms belong to or have
arisen in a different paradigm type, viz. the residual sak -type, and are thus of no relevance to us.
Finally, for the sake of completeness, it should be mentioned that Greek with e.g. its i-stem
gen.sg. -εως < -ηος (< pre-Gr. *- os remodelled from *-e s by application of the lengthened
suffixal vowel of the loc.sg. and the full grade gen.sg. form) and its u-stem gen.sg. -εος (< pre-Gr.
*-e os remodelled from PIE *-e s, cf. Szemerényi (1970: 165)) does not show any instances of
what could be interpreted as o-grade forms in the relevant desinences. Nor does Armenian with e.g.
its i-stem gen.sg. -i < PIE *- os and its u-stem gen.sg. -ow < PIE *- os which should be regarded
either as examples of the type I i- and u-stem inflection or simply as the results of addition of the
productive case endings to the suffix in the form -i-/-u- standardised from the strong cases, cf. Olsen
(1999: 77, 106).
With the relevant data from Indo-Iranian, Balto-Slavic, Italic, Celtic and Anatolian examined and
accounted for, we may now conclude that we could find no unequivocal evidence for suffixal o-
grade forms of the oblique cases of the i- and u-stems in these branches (except for Anatolian
whose o-grade forms belong to or have arisen in a different paradigm type, viz. the residual sak -
type), and we shall now turn our attention to the last of the Indo-European branches in which
apparent, suffixal o-grade forms of the oblique cases of the i- and u-stems seem to occur, viz.
Germanic.

4. Accounting for the Germanic evidence: a selection of previous attempts

Among the Germanic languages, we encounter an array of apparent, suffixal o-grade forms in the
oblique cases of the i- and u-stems, e.g. Goth. i-stem gen.sg. -ais, u-stem gen.sg. -aus, i-stem dat.sg.
-ai, u-stem dat.sg. -au and OE u-stem nom.pl. -a. The communis opinio regarding their origin still

179
seems to be the one presented in e.g. Krahe (1966: 133-134; 1967: 30-33) following, among others,
Brugmann & Delbrück (1911: 135, 156, 175-177, 215):

PIE i-stem gen.sg. *-oĩs > Goth. -ais

PIE i-stem dat.sg. *- > Goth. -ai (as also PWG *-i)
PIE u-stem gen.sg. *-o s > Goth. -aus, ON -ar, OE -a, OS -o, OHG -
PIE u-stem dat.sg. *- > Goth. -au (as also RN -iu, OHG -iu)
PIE u-stem dat.sg. *- > OE -a, OS -o (only Krahe, cf. also Szemerényi (1970: 162))
PIE u-stem voc.sg. *-o > Goth. -au
PIE u-stem nom.pl. *-o es > OE -a (only Krahe, cf. also Szemerényi (1970: 162))59

In this account, Krahe only explains the apparent o-grade forms of the i- and u-stem dat.sg.,
developed from the PIE loc.sg., by PIE e-grade forms. In doing so, he follows Bazell (1937: 1-4)
who, in turn following and elaborating on Streitberg (1896: 245-246), seems to imply the following
phonological development: PIE *- > PG *-ai and PIE *- > PG *-au. The remaining case forms
are still explained by application of PIE o-grade forms. Given the validity of that assumption,
Germanic would, in fact, be the sole Indo-European branch to require unambiguous PIE o-grade
forms in the relevant case forms.
Boutkan (1995: 83-89, 236-257) more or less adopts the communis opinio, represented by
Krahe’s outline in the present article, albeit with a couple of minor exceptions. Firstly, since, in his
interpretation of the Germanic “Auslautgesetze” syllable structure constitutes the decisive factor
for the outcome of the final syllables in the individual Germanic languages (as opposed to the
standard view which is based on the prosodic or moraic nature of the mere vowel of the final
syllable, cf. e.g. the concise account given by Jasanoff (2004: 247-255)), he need not operate with
any difference between acute and circumflex vowels and diphthongs.60 This difference of opinion,
while pivotal to Boutkan’s outline of the Germanic “Auslautgesetze” in general will be of no
relevance to our purpose. Secondly, Boutkan, while accepting the notion that PIE *- > Goth. -ai,
PWG *-i and that PIE *- > Goth. -au, OHG -iu etc., chooses to follow Kortlandt (1990: 6) by

59 Also attested in Old Frisian as -a, cf. e.g. Bremmer (2009: 63). Parallel to the PIE u-stem nom.pl. *-o es > OE -a,
all scholars, incl. Krahe, posit an uncontroversial u-stem nom.pl. PIE *-e es > Goth. -jus, ON -er, OS -i, OHG -i.
60 Boutkan (1995) thus distinguishes between protected (i.e. final vowel followed by consonant) and non-protected
(i.e. vowel in absolute final position) final vowels.

180
positing a different PG outcome in order to account for the coexistence of high- and low-vowel
reflexes from the same PG source in the different Germanic languages, viz. PG *- i and *- u
rather than the Bazell/Krahe variants PG *-ai and *-au. Thirdly, he chooses to elaborate on and
refine an idea by Bazell (1937: 4) that OE -a < PG *-ewiz (not †-iwiz) < PIE *-e es.
In a short but, to our purpose, highly relevant footnote, Rasmussen (1996: 137) suggests that the
Goth. u-stem gen.sg. -aus is developed from PIE *-e s (supposedly through the intermediate stage
of PG *-auz) rather than from *-o s. In this way, as Rasmussen also remarks, no u-stem gen.sg. of
any Indo-European language seems to point to *-o s. By means of analogy to the parallel u-stems,
the Goth. (feminine) i-stem gen.sg. could hardly be reshaped as anything but -ais.
With the addition of Rasmussen’s footnote to the views of Boutkan (elaborating on Bazell), we
are left with a picture that points increasingly towards the elimination of PIE suffixal o-grade forms
in the oblique cases of the i- and u-stems. Nonetheless we can hardly help noticing a somewhat
adverse lack of system in the developments above, i.e. the developments that have been proposed
by previous scholarship – either with or without any disclosure of intent to render the position of
PIE suffixal o-grade forms in the oblique cases of the i- and u-stems superfluous. In addition, none
of the developments posited may account for the Goth. u-stem voc.sg. -au as descending from
anything but PIE *-o .61 Admittedly, most scholars would probably expect a case as marginal as the
voc.sg. to be highly prone to paradigmatic levelling from the more prominent cases; consequently,
this latter objection is of only minor relevance.

5. Accounting for the Germanic evidence: a new attempt

Any new account of the problem presented here should avoid “patchwork solutions” and rather
search for a more general soundlaw or a coherent series of general soundlaws. Consequently, in the

61 I choose to follow Braune & Ebbinghaus (1973: 71) and with them Boutkan (1995: 254-255) in their distributional
analysis of the Goth. u-stem voc.sg. where -au coexists with -u and where free variation seems to prevail. However,
when realising that -au is present 7x in sunau ‘son’ and 1x in magau ‘boy’ whereas -u is only present 1x in sunu and
1x in dauþu apart from 7x in Greek personal names ( Greek voc.sg. -υ), we would have to conclude that -au is
original and that -u has entered the paradigm due to influence from Greek or due to its appearance (apart from the
Greek personal names) in the orthographically unstable Мss. Ambr. A and Ambr. B where, as also in the Gospel of
Luke and in the Ms. Cal., the employment of <u> and <au> vowel forms is seemingly randomised throughout the
entire u-stem paradigm.

181
present study, the following, interconnected series of soundlaws will be suggested as an alternative
to those presented in the previous paragraph:62

PIE *-e > PG *-ai(C)#

PIE *-e > PG *-au(C)#

PIE *- > PG *-ei(C)# (> *-ī #)

PIE *- > PG *-eu(C)#

These soundlaws share the advantage of typological adequacy in relation to the general reductional
and apocopical tendencies of unaccented syllables in Proto-Germanic and, secondarily, in the
individual Germanic languages, cf. e.g. Krahe (1966: 127-134). Thus, both PIE *- > PG *-
ei(C)# and PIE *- > PG *-eu(C)# may be described as shortening by one mora. Even if PIE
*-e > PG *-ai(C)# and PIE *-e > PG *-au(C)# can hardly be cases of true shortening, I
would, while recognising that such a hypothesis is entirely unfalsifiable, tentatively suggest
interpreting the development of PIE *e > *PG a as an underlying reduction through the intermediate
stage of a schwa-like vowel at a very early stage, i.e. before unstressed PG *e started tending
towards developing into *i in various positions. Subsequently, the schwa-like vowel would be
interpreted by the speakers of Proto-Germanic as a. Thus, PIE *-e > pre-PG *-ә > PG *-
63
ai(C)# and, correspondingly, PIE *-e > pre-PG *-ә > PG *-au(C)#.
Table 2 seeks to illustrate how the relevant material may or may not fit with the soundlaws just
proposed. In order to establish the soundlaws, material from the oldest stages of the main Germanic

62 In a narrow sense, only the two first soundlaws offer a solution to the problem of apparent, suffixal o-grade forms of
the oblique cases of the i- and u-stems, but it should be noted that we also need to include the two latter ones in
order to obtain the full picture of the relevant desinences without leaving any form unaccounted for. In other words,
since explaining e.g. RN -iu and OHG -iu from PG *-au (< PIE *-e ) would now seem impossible or at least
unlikely, an alternative explanation must be sought, and, in this case, PG *-eu < PIE *- – a PIE desinence with
cognates in other Indo-European branches – appears to be the straightforward option.
63 Whether a similar development can be upheld for the development of the PIE o/a-vowel diphthongs into Germanic,
i.e. if the development of PIE *-o /*-a and PIE *-o /*-a > PG *-ai and *-au, respectively, could also have passed
through the intermediate stage of pre-PG *-ә and *-ә , respectively, falls outside the scope of the present study.

182
languages, i.e. Gothic, Runic Norse (Proto-Norse),64 Old Norse, Old English, Old Saxon and Old
High German, has been inserted into the table as well.

Table 2: Relevant Germanic material possibl reflect ng I *- and *- :65

PIE PG Goth. RN/ON66 OE OS OHG

i-stem *-ei-s *-aiz qenais vetter-ges b næ/e  dat.sg.  dat.sg.

gen.sg.
Norw. alfer- or  dat.sg.

i-stem *-ei(-ei) *-ai qenai winai b næ/e - -

dat.sg.
brúði

i-stem *- i *-ei > *-ī - - wini D stedi D ensti D

loc.sg.

*-ei *-ai qenai D67 winai D b næ/e D

brúði D

64 I have made no attempt to distinguish between different stages of Runic Norse; consequently, we should apply the
relevant data on our analysis with a considerable amount of caution.
65 A wide array of additional forms has not been listed in the table, viz. the following desinences that have obviously
developed from interparadigmatic analogy, also cf. Krahe (1967: 30-33) and Boutkan (1995: 236-258). For Gothic:
Masculine i-stems in the singular in general ( a-stems), i.e. gen.sg. -is, dat.sg. -a, and voc.sg. -Ø. For Old Norse: i-
stem gen.sg. -s ( a-stems), -ar ( ō- and/or u-stems); i-stem dat.sg. -Ø ( ō-stems), -o ( ō-stems; subsidiary
ending); u-stem gen.sg. -s ( a-stems), u-stem dat.sg. -Ø ( u-stem acc.sg.). For Old English: i-stem gen.sg. -is, -es
( а-stems); i-stem dat.sg. -Ø (only Northumbrian,  old i-stem instr.sg. *-ī or consonant stem instr.sg. *-i?); u-
stem gen.sg. -es ( a-stems), -e ( ō-stems), -Ø (with heavy roots; possibly analogy from the corresponding
dative); u-stem dat.sg. -e ( а- or ō-stems), -Ø (with heavy roots; possibly u-stem instr.sg. used as a dative). For
Old Saxon: i-stem gen.sg. -Ø ( consonant stems), -(i)æs ( ia-stems); i-stem dat.sg. -Ø ( consonant stems),
-(i)æ ( ia-stems), -iu ( iō-stems); u-stem gen.sg. -æs ( a-stems), -(i)æs ( ia-stems); u-stem dat.sg. -æ ( а-
stems), -(i)æ ( iа-stems), -i ( i-stems), -Ø (with heavy roots; possibly u-stem instr.sg. used as a dative). For Old
High German: i-stem gen.sg. -es ( a-stems), -Ø ( consonant stems); i-stem dat.sg. -e ( a-stems), -Ø (
consonant stems); u-stem gen.sg. -es ( a-stems); u-stem dat.sg. -e ( a-stems), -Ø (only in hant; possibly u-stem
instr.sg. used as a dative).
66 In addition maybe Antonsen’s i-stem dat./loc.sg. RN faþai (Charnay Clasp, 550-600 AD). However, this inscription
is poorly understood in almost every regard and can hardly count as evidence for or against the developments
suggested in table 2.
67 The capital D designates the synchronic use of the form as a dative.

183
u-stem *-eu-s *-auz sunaus magoR suna suno fridō
gen.sg.
sonar

u-stem *-eu(-ei) *-au sunau D hakuþo D suna/o/u D suno/u D -

dat.sg.
asau D

u-stem *- u *-eu - mAgiu D - - sitiu/i D

loc.sg.
-mudiu D

syni D

*-eu *-au sunau D hakuþo D suna/o/u D suno/u D -

asau D

u-stem *-u *-u sunu = nom.sg. = nom.sg. = nom.sg. = nom.sg.

voc.sg.
*-eu *-au sunau

u-stem *-eu-es *-ewiz sunjus synir suna suni siti

nom.pl.

5.1. Remarks on the i-stem gen.sg.

While offering much-needed solutions to some of the issues discussed in the present article, my
hypothesis is not entirely seamless and raises new issues to be solved. One such new issue is the RN
i-stem gen.sg. -īz attested in RN ekgudijaungandiz... (Nordhuglo stone, Norway, 425 AD) which
Antonsen (1975: 47) representing the standard interpretation translates as ‘I the priest of
ngandiz’ i.e. RN i-stem gen.sg. -īz < PG *-eis – a form hardly deducible from PIE *-e s if we are
to believe the development of PIE *-e s > PG *-ais posited above; a development PG *-e s < PIE
†- s is even less likely in PIE terms. An alternative and in fact more preferable solution is offered
by Boutkan (1995: 245) who firmly believes RN ungandiz to reflect the ia-stem nom.sg. *-īz (< PG
*-ijaz) i.e. “I the priest ngandijaz”. However thanks to a recent e-mail correspondence with
Roland Schuhmann, I now regard the case form RN -īz as resulting from the i-stem nom.sg. -iz
(rather than i-stem gen.sg. or ia-stem nom.sg. †-īz) seeing that, in contemporaneous RN forms such
as RN stainawarijaz, the alleged contraction of PG *-ijaz > RN -īz in post-unstressed position has
not taken place.
Puzzling high-vowel reflexes of the i-stem gen.sg. are not found only within Runic Norse. Also
the PWG reflex *-ī poses at least a couple of issues to be accounted for. The main issue strongly
resembles that of the RN form above, viz. that PWG *-ī would imply an i-stem gen.sg. PG *-īz <

184
PG *-eis < (non-existing) PIE †- s. The second issue is how to account for the parallel OE i-stem
gen.sg. -æ/-e which seemingly reflects the expected form PG *-aiz < PIE *-e s. Two mutually
exclusive strategies may be applied. We may choose to follow either the explanation offered in the
present article that PWG *-ī (< *-iz), paralleling the situation of the ō-stems, had spread
analogically from the i-stem dat.sg. PWG *-ī before the latter was regularly shortened or the
proposal outlined by Boutkan (1995: 245), following Kortlandt (1990: 6), that PG *-aiz > pre-PWG
*-æi(z) > PWG *-ī, implying that the OE i-stem gen.sg. -æ/-e does not represent a direct
development from PG *-aiz but rather has been taken over from the ō-stems by means of
interparadigmatic analogy.

5.2. Remarks on the i- and u-stem dat./loc.sg.

What would most likely disturb many scholars of Indo-European linguistics is the assumption,
implied by my hypothesis, of an endingless i-stem loc.sg. PIE *-e and, correspondingly, of a u-
stem loc.sg. *-e , i.e. forms without the hic-et-nunc particle PIE *-i. Even if endingless loc.sg.
forms are well attested within other paradigms, cf. e.g. Skt. n-stem loc.sg. r jan ‘king’ beside r jan-
i, they are normally not reconstructed for the PIE i- and u-stems which does not necessarily imply,
however, that it is fat-fetched to do so. Such forms might actually be attested in Indo-Iranian, cf.
Brugmann & Delbrück (1911: 176-177) and, in more detail, Beekes (1985: 112), e.g. Av. rū te ‘to
say’ Skt. vásto ‘by illumination’ (infrequent ending) OPers. g ϑav- ‘on the square’ etc. For
various reasons including a number of sandhi mechanisms, however, Wackernagel & Debrunner
(1929: 155) chooses to disregard the existence of such PIE short-diphthong loc.sg. forms in Indo-
Iranian. With the dubious relevance of the short-diphthong loc.sg. forms borne in mind, the dat.sg.
may offer a more attractive alternative. Granted that an i-stem dat.sg. PIE *-e developed from PIE
*-e -e by haplology,68 we could easily assume the analogical creation of a corresponding u-stem
dat.sg. PIE *-e .
However, though disregarded by Wackernagel & Debrunner (1929: 155), the idea of an
endingless loc.sg. should perhaps not be abandoned quite yet seeing that, in fact, a u-stem loc.sg.
PIE *-e might be exactly what underlies Gr. ἄνευ prep. ‘without far from’ to which not only a
desinential but also a perfect, lexical cognate may be found in PG *enau (< PIE *-e ) and * n

68 As has also been suggested for e.g. Lat. -ī, Osc. -eí, and OCS -i, cf. Brugmann & Delbrück (1911: 170-171) and
Szemerényi (1970: 162). Sihler (1995: 316), however, posits a regular, phonological development PIE *-e -e >
*-ee > *- > -ei/- > Lat. -ī.

185
prep. ‘without’ as reflected in OHG aano, ano prep. ‘id.’ (with variants ana, ane, an and ultimately
ânu < PG * n ) and OS no prep. ‘id.’, both from PG * n-au, cf. Lloyd & Springer (1988: 289-
290). Beekes (2010: 102) rejects any etymological connection between the Greek and the Germanic
forms and prefers to reconstruct Gr. ἄνευ as a u-stem loc.sg. *sn̥(H)-eu to PIE *sen(H)- ‘without’
cf. also Skt. sanutár- prep. ‘away off aside’ and Lat. sine prep. ‘without’ based on his
disinclination towards Eichner’s Law. Consequently in Beekes’ view even if Gr. ἄνευ could
formally reflect PIE * 2n-e , PG * n a < PIE * n e could never reflect PIE *h2 n-e - with lack
of laryngeal colouring of a following long vowel as suggested by Nikolaev (2007: 165). Whether or
not we accept Nikolaev’s explanation through Eichner’s Law we would have to agree that the
Goth. by-form inu prep. ‘without’ (< PG *enu or *inu) is hardly explicable from any root
constellation containing PIE *h2. Kroonen (2013: 118) does provide a solution to the Goth. inu,
though, viz. the application of the heavily debated process of pretonic shortening also known as
Dybo’s Law. As such Kroonen would reconstruct PIE *h2 n-ú- > post-PIE * nú > PG *enu > Goth.
inu, and PIE *h2 n-e - > PG * na - > PWG * na -69 > OHG aano etc. Alternatively, inu could be
a mere Gothic shortening of PG * n in a weakly stressed, i.e. prepositional, form.

5.3. Remarks on the u-stem nom.pl.

Admittedly, the u-stem nom.pl. does not contain a word-final diphthong. The reason for its
inclusion in table 2 in spite of this is the dual purpose of the present study, viz. (1) to eliminate or
render superfluous all previously posited instances of suffixal o-grade vowels in the oblique cases
of i- and u-stems, and (2) to provide evidence for a new coherent series of soundlaws affecting
word-final diphthongs. Whereas the u-stem nom.pl. with its word-internal diphthong is of no
relevance to our purpose no. (2), it certainly does meet the requirements for being relevant to our
purpose no. (1).70 As previously mentioned, only the OE (and Old Frisian) u-stem nom.pl. -a, which
has traditionally been reconstructed as PIE *-o es, may cause us problems in our attempt to satisfy

69 Or in Kroonen’s (2013: 118) terms NWGm. * ne seeing that such a form would also account perfectly well for
ON án, ón prep. ‘without’.
70 The nom.pl. is a strong case; not a weak or an oblique one. Consequently, we could reasonably argue that it also falls
outside the scope of purpose no. (1), viz. to eliminate or render superfluous all previously posited instances of
suffixal o-grade vowels in the oblique cases of i- and u-stems. It is included here nonetheless because it contains the
same suffixal PIE *-o - as the weak or oblique cases.

186
our purpose no. (1); the remaining Germanic forms may be derived straightforwardly from the
expected form, i.e. PIE *-e es. The OE form may be accounted for in a number of ways.
Firstly, we might consider regarding it as the result of an analogical process in which the original
u-stem nom.pl. was replaced with the corresponding form of other stems e.g. the ō-stems (West
Saxon and Late Kentish -a) or the a-stems if we dare run the risk of operating with either an
otherwise unattested a-stem nom.pl. OE *-a < PG *- z besides attested -as/-æs or an acc.pl. OE *-a
< PG *-anz as attested in OHG, OS and Old Low Franconian (but not in OE where the nom.pl.
-as/-æs has replaced the old acc.pl.). As for the a-stem nom.pl., another and far more serious issue is
that a reconstruction along the lines of PG *- z for the a-stem nom.pl. might not be valid at all.
According to Boutkan (1995: 187-191), all the attested forms could and should be explained from
PG *- s-ez (> Goth. - s, OE -as/-æs, OS -as/-os) alternating with PG *- z-ez (> Goth. - s, ON -ar,
OFris. -ar (especially in the area around Emsigoland)).71 As such, no actual foundation for that
specific analogy remains.
Providing us with a second alternative, Bammesberger (1985: 366-370) suggests that OE -a
continues the old a-stem nom.du. PG *-au < PIE *- ). According to Bammesberger, the
motivation for such a substitution would be found in the fact that the original u-stem nom.pl. would
virtually disappear in Old English if the soundlaws ran their course, i.e. traditionally expected PG
*-iwiz > pre-OE *-ju > OE -Øi (geminating effect of Umlaut-causing *-j-; subsequent loss of -u
after heavy syllables, e.g. pre-OE *sunju > OE †synn ‘sons’). Whether or not such a suggestion
might seem attractive to the individual scholar, we would have to admit that, by accepting it, we
would run a risk identical to that of one of the first alternatives, viz. that of operating with an
otherwise unattested form.
In my view Boutkan’s (1995: 83-89) suggestion that OE -a is the result of a regular,
phonological development therefore seems far more appealing. As such, PIE *-e es > PG *-ewiz
(rather than PG †-iwiz, as usually assumed) > PWG *-ew [vel sim.] > OE -a, OS -i, OHG -i.

5.4. Diphthongal desinences reflecting PIE *- *- and *- *- or similar constellations

Although other diphthongal desinences (reflecting PIE *- *- and *- *- or similar

constellations) should ideally be left out of consideration here as not directly pertinent to the present
study, a list of expected correspondences will nevertheless be given in table 3 so as to illustrate in
particular that, when we apply our new soundlaws on the Germanic material, a problematic

71 OHG - /-a may be explained as the acc.pl. form, cf. Boutkan (1995: 191).

187
difference arises between the outcome of “original” PG *-ai (< PIE *-o /*-a ) and “secondary” PG
*-ai (< PIE *-e ).

Table : er an c ater al reflect ng I *- *- and *- *- :

PIE PG Goth. RN/ON OE OS OHG
o-stem *-ōi *-ai - hanhai dægæ/e daga/e tage
dat.sg. ride
degi
o-stem *-oi *-æ ? ūta uti ūte ūta/e ūz(z)e
loc.sg. (daga D) úti
o-stem *-ō(u) *-au ahtau átta eahta ahto ahto
nom.du.
o-stem *-oi *-æ ? blindai72 -arjoste-z blinde blinda/e blinte
nom.pl. blindi-r
(strong adj.)
ā-stem *-eh2-ei *-ai gibai - giefæ/e geba/e -
dat.sg.
ipv.3.sg. *-e-tu  *-adau ? nimadau73 - - - -
ipv.3.pl. *-o-ntu  *-andau ? habandau - - - -
N
opt.1.sg. *-oih1-m *-ai (nimau) (nema) nime nima/e neme
opt.2.sg. *-oih1-s *-aiz nimais nemir nime (nima/es) (nem s)
opt.3.sg. *-oih1-t *-ai(þ) nimai nemi nime nima/e neme
opt.3.pl. *-oih1-nt *-ain nimain-a nemi nimen nima/en nem n
pass. *--oi *--æ ? haitada haite hātte - -
heiti
opt.pass. ? *--au ? haitaidau - - - -

72 Goth. -ai is found in this desinence rather than the expected †-a due to analogical influence from stressed final
diphthongs of monosyllabic words; more precisely from the corresponding forms of the demonstrative pronoun
represented, in this case, by Goth. m.nom.pl. þai ‘they’.
73 In Gothic, we encounter a series of enigmatic verb forms ending in -au, viz. the imperative 3.sg. -adau, the
imperative 3.pl. -andau, and the optative passive (1./3.sg. -aidau, 2.sg. -aizau, pl. -aindau). As for the source of the
imperative forms, Boutkan (1995: 326-327, 355-356) has ingeniously suggested a combination, or contamination, of
PIE *-et d (fut.ipv.) and PIE *-(n)tu (ipv.3.sg./pl.). So far, however, the exact prehistory of the forms of the optative
passive remains enigmatic. In any case, these forms are of no relevance to the present study and will therefore not be
discussed into further detail.

188
In protected position, i.e. when followed by a consonant, the two different PG *-ai’s seem to
produce the same outcome in the individual Germanic languages, viz. e.g. Goth. -aiC, ON -eC, OE
-æC/eC, OS -æC, and OHG - . When the PG *-ai’s are in absolute final position, however, the
clarity of the developmental chain rapidly decreases. In other words, the two *-ai’s produce
different outcomes in the individual Germanic languages, viz. Goth. -a, ON -e, OE -e, OS -æ, OHG
-e (for “original” PG *-ai < PIE *-o /*-a ) vs. Goth. -ai, RN -ai, ON -e, OE -æ/-e (for “secondary”
PG *-ai < PIE *-e ). Due to the divergent quality of the resulting vowels of the “original” PG *-a
in absolute final position it has tentatively been denoted as ‘*-æ ?’ in table 3. In my view, this
situation can be understood in only one way viz. that the “original” PG *-ai had developed into *-æ
vel sim. in absolute final position before PIE *-e developed into “secondary” PG *-a ; otherwise
even “secondary” PG *-a would have developed into *-æ vel sim. in absolute final position.74 We
may thus reasonably – by adhering to Boutkan (1995: 468) for at least the more commonly accepted
stage (1) – state the following relative chronology of our delicate ‘PG *-ai problem’:

(1) PG *-aiC# > Goth. -aiC#

(1) PG *-ai# > Goth. -a# (hence the tentative notion of PG *-æ# rather than *-ai#)

(2) PIE/pre-PG *-eiC# > PG *-aiC# > Goth. -aiC#

(2) PIE/pre-PG *-ei# > PG *-ai# > Goth. -ai#

6. Conclusion
By proposing a series of four Germanic soundlaws (illustrated in table 4) pertaining to PIE word-
final e-vowel diphthongs, the present article offers a coherent explanation of the Germanic evidence
of alleged o-grade forms of the gen.sg. and loc.sg. of PIE i-stems and the gen.sg., loc.sg. and voc.sg.
of the PIE u-stems; and by adopting Boutkan’s explanation of the alleged o-grade form PIE *-o es
of the u-stem nom.pl. (as seen in OE -a), I have found alternative explanations for all Germanic
evidence of alleged o-grade forms in the oblique cases of PIE i- and u-stems. With previous

74 Guus Kroonen (p.c.) has perceptively pointed to the fact that, even if the development of PIE *-e > PG *-a be
posterior to that of “original” PG *-ai > *-æ vel sim. in absolute final position, it must predate the apocope of
PIE/pre-PG *e in the imperative 2.sg. of the Germanic class I weak verbs, cf. e.g. Goth. -ei. Otherwise, this form,
too would have developed into †-ai.

189
scholarship having offered alternative explanations of the evidence from the remaining Indo-
European branches, we may therefore state that the very notion of o-grade forms in the oblique
cases of PIE i- and u-stems (as also in the strong cases of the voc.sg. and nom.pl.) has now been
rendered superfluous.
In addition, I have suggested that the development of at least PIE *-e > PG *-ai must postdate
that of original (pre-)PG *-ai# (from PIE *-o /-a ) > PG *-æ# vel sim., cf. also table 4, since these
two outcomes would otherwise have coalesced.

Table 4: Summary of the phonological development from PIE to Germanic:

PIE PG Goth. RN/ON OE OS OHG
*-ei / (C) *-ai -ai -ai/-e -æ/-e - -
-i
*-eu / (C) *-au -au -au/-o -a > -o/-u -o/-u -ō
-a
*- i / (C) *-ei > *-ī - - -i -i -i
*- u / (C) *-eu - -iu - - -iu > -i
-i
*-oi / (C) *-aiC# -aiC# -ai/-e -e -a/-e - C
*-æ# ? -a# -i -e#
*-ou / (C) - - - - - -
*-ōi / (C) *-ai -ai -e -æ/-e -a/-e -e
-i
*-ōu / (C)# *-au -au -a -a -o -o

It should be noted, though, that previous scholarship has offered partial alternatives to the Germanic
evidence, as well, cf. e.g. Bazell (1937: 1-4) and Rasmussen (1996: 137). To my knowledge,
however, my account of the problem is the first to offer a coherent solution to all the evidence and
also the first to operate mainly with regular soundlaws rather than analogical processes. As such,
every Germanic form relevant to the present study – with the sole exception of either the PWG i-
stem gen.sg.*-ī or the OE i-stem gen.sg. -æ/-e – can be explained by the mere application of regular
soundlaws.

190
3. Conclusion

As I stressed in the beginning of my thesis, its primary aim and purpose is to identify archaisms in
Germanic and to separate these archaisms from innovations which have been disturbed by
analogical processes, thereby aiding the scholarly community in its quest to reconstruct and obtain
greater knowledge about earlier linguistic stages. In other words, only by knowing what elements of
Germanic can be discarded as innovations created by means of analogy will the scholarly
community be capable of inferring information from Germanic for the understanding of earlier
linguistic stages such as, above all, Proto-Indo-European.

3.1. Archaisms and innovations in nominal derivational morphology

I have conducted two studies on archaisms and innovations in the field of Germanic nominal
derivation where phonological developments of unstressed syllables have led to a collapse and a
subsequent rearrangement of the entire system, thereby making it virtually impossible for the
scholarly community to identify the important archaisms. These two studies on Germanic nominal
derivation and inflection centre on root nouns and primary i-stem nouns.
In Ph.D. article no. 1, I have argued that root nouns in Proto-Germanic and the individual
Germanic languages may be attributed to three chronologically defined layers, the first of which
consists of root nouns inherited from Proto-Indo-European. Building on an idea originally
developed by Nielsen Whitehead (2010; 2013), I have further demonstrated that we may recognise
the archaic, inherited root nouns from their phonotactic characteristics. With the exception of a
mere handful of forms which can all be otherwise accounted for, inherited root nouns, which were
originally ablauting in Proto-Indo-European, generalise their radical ablaut grade so as for roots of
the structure (C)CVRC to level in favour of the zero grade originally pertaining to the weak stem
and for roots of the structures (C)CVT(C), (C)CVH(C) and CVC to level in favour of a fuller ablaut
grade originally pertaining to the strong stem. Regardless of the radical phonotactics, however, a
true PIE a-vowel always remains and is never reduced into the zero grade.
Conversely, the second and third layers consist of root nouns that have entered this inflectional
class at a later stage and are thus to be regarded as innovations. Lexical borrowings and nouns from
other inflectional classes reanalysed as root nouns in parts of the Germanic dialect continuum
constitute the second layer, whereas the huge amount of North Germanic root nouns originating

191
from other inflectional classes belong to the third layer which may actually be regarded as a mere
North Germanic continuation and further development of the second layer.
As for the Germanic primary i-stem nouns treated in Ph.D. article no. 2, I have ventured to
demonstrate that, even though i-stem nouns functioning as verbal abstracts are synchronically and
productively formed from a stem displaying the same ablaut grade as the preterite participle of
related strong verbs, truly archaic i-stems display only radical o- and zero grade. Moreover, these
two types, i.e. the zero grade type and the o-grade type, were originally identical. As demonstrated
by Rasmussen (1989: 158-175) for the Proto-Indo-European feminine eh2-stems, i.e. the Germanic
ō-stems, the radical phonotactics constitutes the determining factor for when to expect zero grade
and when to expect o-grade. When this system was transferred to Germanic, the original
distribution of o- and zero grade was abandoned in favour of a new ablaut distribution dependent on
the ablaut system of the strong verbs, and the old system is only recognisable through a handful of
i-stems displaying aberrant ablaut grades in comparison with the strong verbs pertaining to them.
This new system together with, i.a., the reanalysis of old s-stems as i-stems has given rise to a
wide array of possible Germanic ablaut grades in the root syllable, i.e. zero grades, o-grades, full
grades and lengthened grades, but anyone of these not adhering to the distributional rule of o- and
zero grade described above is to be regarded as an innovation and can thus safely be discarded.

3.2. Support from Germanic historical phonology

Every historical and comparative morphological analysis relies on the assumption of a set of regular
and recurring phonological correspondences between the languages studied. Hence follows that, if
these correspondences have been suggested on an erroneous basis, any morphological analysis
building on them runs the risk of being severely flawed, as well. Thus, the importance of
phonological regularity cannot be underestimated.
For this reason, I have included two articles suggesting new Germanic sound changes in my
thesis. In Ph.d. article no. 3 I have demonstrated that, contrary to common belief, PG *#ai- stands a
good chance of being the regular outcome of PIE *#h2i- just as PG *#au- is likely to be the regular
outcome of PIE *#h2u-. At least that is what is clearly suggested by the statistical evidence of a
double-digit number of possible or likely candidates of PIE *#h2i-/*#h2u- > PG *#ai-/*#au- as
opposed to no possible or likely candidates of PIE *#h2i-/*#h2u- > PG *#i-/*#u-. As for the
sequences of initial PIE *h1 and possibly also *h3 followed by PIE *i or *u, there is no compelling
reason for discarding the communis opinio of these sequences yielding PG *#i- and *#u-,
respectively. Furthermore, if we can draw parallels to Germanic from the developments seen in

192
Greek and English, viz. that any example of PG *#au- followed by a labial consonant would result
in PG *#u-, it would seem that the conclusions presented here remain valid regardless of the
pedigree of PG *ubila- adj. ‘evil bad’ *uba prep. ‘under; above’ and *ufna- m. ‘oven’ which are
often seen reconstructed with initial PIE *h2 in the scholarly literature.
The proposed sound change of PIE *#h2i- > PG *#ai- sheds new light on the etymology of PG
*aik- f. ‘oak’. If not a lexical borrowing from some unknown source and if not containing an
original (post-)PIE *a, PG *aik- would normally be analysed etymologically as PIE *h2e ĝ- or
*h2e ĝ-, thereby violating the distributional rules governing the assignment of radical ablaut grade
to root nouns according to the structure of their roots, but with the application of this new regular
sound change of PIE *#h2i- > PG *#ai-, we are capable of reconstructing PG *aik- as PIE *h2 ĝ-
which conforms perfectly well to the distributional rules outlined above.
The last article, i.e. Ph.d.-article no. 4, is concerned with the linguistic prehistory of the
Germanic i- and u-stem desinences, thereby potentially adding to our understanding of analogical
and other processes operating in the i-stems as well as in the u-stems . By proposing a series of four
Germanic sound changes pertaining to PIE word-final e-vowel diphthongs, viz. PIE *-e > PG
*-ai(C)#, PIE *-e > PG *-au(C)#, PIE *- > PG *-ei(C)# > PG *-ī and PIE *-
> PG *-eu(C)#, I have offered a coherent explanation for the Germanic evidence of alleged o-grade
forms of the gen.sg. and loc.sg. of PIE i-stems and the gen.sg., loc.sg. and voc.sg. of the PIE u-
stems. Furthermore by adopting Boutkan’s (1995: 83-89) explanation for the alleged o-grade form
PIE *-o es of the u-stem nom.pl. as attested in OE -a, I believe to have found alternative
explanations for all Germanic evidence of alleged o-grade forms in the oblique cases of PIE i- and
u-stems. Consequently, with only one exception, we can now explain every desinence of the
Germanic i- and u-stems by the mere application of processes of regular sound change. In other
words, almost every such desinence must now be regarded as an archaism.

193
References

Adams, Douglas Q. 1988. Tocharian historical phonology and morphology (American Oriental
Society 71). New Haven, CT: American Oriental Society.
Adams, Douglas Q. 1999. A dictionary of Tocharian B (Leiden Studies in Indo-European 10).
Amsterdam: Rodopi.
Andersen, Harry. 1962. Oldnordisk grammatik. 3. rev. udg., 2. opl. Kbh.: J.H. Schultz.
Antonsen, Elmer H. 1975. A concise grammar of the older Runic inscriptions. Tübingen: Max
Niemeyer Verlag.
Bammesberger, Alfred. 1985. Die Endung für Nom. Akk. Pl. bei altenglischen u-Stämmen. Anglia
103. 365-370.
Bammesberger, Alfred. 1990. Die Morphologie des urgermanischen Nomens (Untersuchungen zur
vergleichenden Grammatik der germanischen Sprachen 2). Heidelberg: Carl Winter
Universitätsverlag.
Bammesberger, Alfred. 2007. The etymology of Germanic *idis-. NOWELE 52. 81-89.
Bazell, Charles E. 1937. IE final unaccented in Germanic. The Journal of English and Germanic
Philology 36/1. 1-9.
Beekes, Robert S.P. 1969. The development of Proto-Indo-European laryngeals in Greek. The
Hague: Mouton.
Beekes, Robert S.P. 1985. The origins of the Indo-European nominal inflection (Innsbrucker
Beiträge zur Sprachwissenschaft 46). Innsbruck: Institut für Sprachwissenschaft der Universität
Innsbruck.
Beekes, Robert S.P. 1988. Laryngeal developments: A survey. In Alfred Bammesberger (ed.), Die
Laryngaltheorie und die Rekonstruktion des indogermanischen Laut- und Formensystems, 59-
105. Heidelberg: Carl Winter Universitätsverlag.
Beekes, Robert S.P. 1995. Comparative Indo-European linguistics. An introduction. Amsterdam:
John Benjamins Publishing Company.
Beekes, Robert S.P. 2010. Etymological dictionary of Greek (Leiden Indo-European Etymological
Dictionary Series 10/1-2). Leiden: Brill.
Benveniste, Émile. 1935. Origines de la formation des noms en indo-européen. Paris: Librairie
Audrien-Maisonneuve.
Bjorvand, Harald & Fredrik O. Lindeman. 2000. Våre arveord. Oslo: Universitetsforlaget.

194
Bloomfield, Leonard. 1933. Language. New York: Henry Holt.
Boutkan, Dirk. 1995. The German c ‘A sla tgesetze’ (Leiden Studies in Indo-European 4).
Amsterdam: Rodopi.
Boutkan, Dirk & Sjoerd M. Siebinga. 2005. Old Frisian etymological dictionary (Leiden Indo-
European Etymological Dictionary Series 1). Leiden: Brill.
Braune, Wilhelm & Ernst A. Ebbinghaus. 1973. Gotische Grammatik. 17. Aufl. Tübingen: Max
Niemeyer Verlag.
Bremmer, Rolf H. 2009. An introduction to Old Frisian. Amsterdam/Philadelphia: John Benjamins.
Brugmann, Karl. 1901. Homerisch μενοινάω und gotisch briggan, zwei Fälle von
Wurzelangleichung. Indogermanische Forschungen 12. 150-158.
Brugmann, Karl & Berthold Delbrück. 1906. Grundriß der vergleichenden Grammatik der
indogermanischen Sprache II/1. 2. Bearb. Strassburg: Karl J. Trübner.
Brugmann, Karl & Berthold Delbrück. 1911. Grundriss der vergleichenden Grammatik der
indogermanischen Sprachen II/2. 2. Bearb. Strassburg: Karl J. Trübner.
Brøndum-Nielsen, Johs. 1935. Gammeldansk Grammatik 3. Kbh.: J.H. Schultz.
Buck, Carl D. 1904. A grammar of Oscan and Umbrian. Boston: Ginn & Company.
Campbell, Lyle. 2013. Historical linguistics. An introduction. 3. ed. Edinburgh: Edinburgh
University Press.
Casaretto, Antje. 2004. Nominale Wortbildung der gotischen Sprache. Die Derivation der
Substantive. Heidelberg: Universitätsverlag Winter.
Chantraine, Pierre. 1933. La formation des noms en grec ancien (Collection linguistique publiée par
la Société de Linguistique de Paris 39). Paris: Librairie Ancienne Honoré Champion.
Clackson, James. 2007. Indo-European linguistics. An introduction. Cambridge: Cambridge
University Press.
Cohen, Paul S. & Adam Hyllested. 2012. A new sound law of PIE: Initial **h3 > *h2 . In
Benedicte Nielsen Whitehead et al. (eds.), The Sound of Indo-European: Phonetics, phonemics,
and morphophonemics (Copenhagen Studies in Indo-European 4), 53-72. Copenhagen: Museum
Tusculanum Press.
Cowgill, Warren. 1965. Evidence in Greek. In Werner Winter (ed.), Evidence for laryngeals, 142-
180. The Hague: Mouton & Co.
Darms, Georges. 1978. Sc w er nd Sc wager Ha n nd H n. D e rdd -Ableitung im
Germanischen. München: R. Kitzinger.

195
Darski, Józef. 2004. Linguistisches Analysemodell. Definitionen grundlegender grammatischer
Begriffe. 2. völlig neubearb. und ergänzte Aufl. Poznań: Wydawnictwo Naukowe AM.
Derksen, Rick H. 2003. Slavic *jь-. Studies in Slavic and General Linguistics 30. 97-105.
D’iakonov Igor M. 1985. On the original home of the speakers of Indo-European. The Journal of
Indo-European Studies 13(1). 92-174.
Eichner, Heiner. 1973. Die Etymologie von heth. mehur. Münchener Studien zur
Sprachwissenschaft 31. 53-107.
Eichner, Heiner & Robert Nedoma. 2000. Die Merseburger Zaubersprüche: Philologische und
sprachwissenschaftliche Probleme aus heutiger Sicht. Die Sprache 42/1-2. 1-195.
Forssman, Bernhard. 1982-1983. Buchbesprechungen: Peters, Martin: Untersuchungen zur
Vertretung der indogermanischen Laryngale im Griechischen, = Österreichische Akademie der
Wissenschaften, Philosophisch-historische Klasse, Sitzungsberichte, 377. Band. Wien 1980,
Verlag der Österreichischen Akademie der Wissenschaften. X, 346 S., brosch. 80,- DM.
Zeitschrift für vergleichende Sprachforschung 96. 290-292.
Fortson, Benjamin W. 2004. Indo-European language and culture. An introduction. Oxford:
Blackwell Publishing.
Friedrich, Paul. 1970. Proto-Indo-European trees. Chicago: University of Chicago Press.
Frisk, Hjalmar. 1944. Etyma Armeniaca. Göteborgs Högskolas Årsskrift 50. 5-32.
Frisk, Hjalmar. 1960. Griechisches etymologisches Wörterbuch I. Heidelberg: Carl Winter
Universitätsverlag.
Frisk, Hjalmar. 1963-1966. Griechisches etymologisches Wörterbuch II. Heidelberg: Carl Winter
Universitätsverlag.
Fulk, Robert D. 1988. PIE *ə in Germanic unstressed syllables. In Alfred Bammesberger (ed.), Die
Laryngaltheorie und die Rekonstruktion des indogermanischen Laut- und Formensystems, 153-
177. Heidelberg: Carl Winter Universitätsverlag.
Gąsiorowski Piotr. 2012. PIE *-sr- in the context of Verner’s Law. In Benedicte Nielsen Whitehead
et al. (eds.), The Sound of Indo-European: Phonetics, phonemics, and morphophonemics
(Copenhagen Studies in Indo-European 4), 53-72. Copenhagen: Museum Tusculanum Press.
García Ramón, José Luis. 1999. Zur Bedeutung indogermanischer Verbalwurzeln: *h2nek̂-
‘erreichen reichen bis’ *h1nek̂- ‘erhalten (weg)nehmen’. In Jürgen Habisreitinger et al. (eds.),
Gering und doch von Herzen. 25 indogermanische Beiträge. Bernhard Forssman zum 65.
Geburtstag. Wiesbaden: Reichert Verlag.

196
Givón, T. 2000. Internal reconstruction: as method, as theory. In Spike Gildea (ed.), Reconstructing
grammar. Comparative linguistics and grammaticalization (Typological Studies in Language
43), 107-159. Amsterdam: John Benjamins Publishing Company.
Godfrey, John J. 1967. Sir William Jones and Père Coeurdoux: a philological footnote. Journal of
the American Oriental Society 87. 57-59.
Griepentrog, Wolfgang. 1995. Die Wurzelnomina des Germanischen und ihre Vorgeschichte
(Innsbrucker Beiträge zur Sprachwissenschaft 82). Innsbruck: Institut für Sprachen und
Literaturen der Universität Innsbruck.
Grimm (1844 [1865]) = Grimm, Jacob G. 1844. Über zwei entdeckte Gedichte aus der Zeit des
deutschen Heidenthums. In Philologische und historische Abhandlungen der Königliche
Akademie der Wissenschaften aus dem Jahre 1842, 1-24. Berlin: Druckerei der Königlichen
Akademie der Wissenschaften. [Reprinted in: Grimm, Jacob G. 1865. Kleinere Schriften 2, 1-29.
Berlin.]
Güntert, Hermann. 1914. Über Reimwortbildungen im Arischen und Altgriechischen. Eine
Sprachwissenschaftliche Untersuchung (Indogermanische Bibliothek. 3. Abt.: Untersuchungen
1). Heidelberg: Carl Winter’s niversitätsbuchhandlung.
Hammerich, Louis L. 1948. Laryngeal before sonant. København: Ejnar Munksgaard.
Hamp, Eric P. 1978. Indo-European ‘duck‘. Zeitschrift für vergleichende Sprachwissenschaft 92.
29-31.
Hamp, Eric P. 1982. Western Indo-European notes. Indogermanische Forschungen 87. 76-81.
Hansen, Bjarne S.S. 2007. Protoindoeuropæiske laryngaler i germansk: En oversigt over
laryngalernes udvikling fra protoindoeuropæisk til urgermansk. University of Copenhagen price
dissertation.
Hansen, Bjarne S.S. forthc. 2015. Review of Michael L. Kotin. 2012. Gotisch. Im (diakronischen
und typologischen) Vergleich (Sprache – Literatur und Geschichte. Studien zur
Linguistik/Germanistik 41). Heidelberg: Universitätsverlag Winter. NOWELE 68/1. 121-128.
Harðarson Jón Axel. 1987. Das uridg. Wort für “Frau”. Münchener Studien zur Sprachwissenschaft
48. 115-137.
Heidermanns, Frank. 1993. Etymologisches Wörterbuch der germanischen Primäradjektive (Studia
Linguistica Germanica 33). Berlin: Walter de Gruyter.
Van Helten, W.L. 1910-1911. Grammatisches. Beiträge zur Geschichte der deutschen Sprache und
Literatur 36. 435-515.

197
Hill, Eugen. 2009. Das starke Präteritum der Klasse VII in den nord- und westgermanischen
Sprachen. International Journal of Diachronic Linguistics and Linguistic Reconstruction 6(1-2).
49-123, 173-213.
Hinderling, Robert. 1967. Studien zu den starken Verbalabstrakta des Germanischen (Quellen und
Forschungen zur Sprach- und Kulturgeschichte der germanischen Völker, Neue Folge 24 (148)).
Berlin: Walter de Gruyter & Co.
Hirt, Hermann A. 1900. Der indogermanische Ablaut, vornehmlich in seinem Verhältnis zur
Betonung. Strassburg: Verlag von Karl J. Trübner.
Hirt, Hermann A. 1929. Indogermanische Grammatik 5. Heidelberg: Carl Winter.
Hoffmann, Karl & Bernhard Forssman. 1996. Avestische Laut- und Flexionslehre (Innsbrucker
Beiträge zur Sprachwissenschaft 84). Innsbruck: Institut für Sprachwissenschaft der Universität
Innsbruck.
Hollifield, Patrick H. 1979. Final * in monosyllables in North and West Germanic. Die Sprache
25(1). 54-56.
Hollifield, Patrick H. 1984. Raising in unaccented syllables in Germanic. Die Sprache 30. 29-72.
Holthausen, Ferdinand. 1974. Altenglisches etymologisches Wörterbuch. 3. unveränderte Aufl.
Heidelberg: Carl Winter Universitätsverlag.
Hyllested, Adam. 2008. Saami loanwords in Old Norse. In Hans Frede Nielsen (ed.), Early and pre-
historic language development in North-Western Europe (NOWELE 55/56), 131-146. Odense:
University of Southern Denmark Press.
Hyllested, Adam. 2012. Hagl, tavl og skagle. Endnu en undtagelse til den germanske
lydforskydning? In Erik W. Hansen et al. (eds.), Ældre germansk sproghistorie. Et uformelt
minsymposium (Mindre skrifter udgivet af Center for Middelalderstudier, Syddansk Universitet,
Odense 29), 5-12. Odense: Center for Middelalderstudier, Syddansk Universitet, Odense.
Hyllested, Adam & Paul Cohen. 2007. Monophthong for expected -diphthong in Greek. In Coulter
H. George et al. (eds.), Greek and Latin from an Indo-European perspective (Cambridge
Classical Journal: Proceedings of the Cambridge Philological Society, supplementary vol. 32), 9-
18. Cambridge: Cambridge University Press.
IEW = Pokorny, Julius. 2005. Indogermanisches etymologisches Wörterbuch I. 5. Aufl. Tübingen:
A. Fraencke.
Jasanoff, Jay. H. 1989. Old Irish bé ‘woman’. Ériu 40. 135-141.

198
Jasanoff, Jay H. 2004. Acute vs. circumflex: Some notes on PIE and Post-PIE prosodic phonology.
In Adam Hyllested et al. (eds.), Per aspera ad asteriscos. Studia Indogermanica in honorem Jens
Elmegård Rasmussen sexagenarii Idibus Martiis anno MMMIV (Innsbrucker Beiträge zur
Sprachwissenschaft 112), 247-255. Innsbruck: Institut für Sprachen und Literaturen der
Universität Innsbruck.
Johansson, Karl F. 1891. Gotische Etymologieen. Beiträge zur Geschichte der deutschen Sprache
und Literatur 15. 223-242.
Joseph, Brian D. 1975. Laryngeal before i/u in Greek: the role of morphology in diachronic change.
In Robin E. Grossman et al. (eds.), Papers from the 11th Regional Meeting of the Chicago
Linguistic Society, 319-328. Chicago: University of Chicago Press.
Kiehnle, Catharina. 1979. ed sc k nd k vak . Wortgeschichtliche und exegetische
Untersuchungen (Alt- und Neu-Indische Studien 21). Wiesbaden: Franz Steiner Verlag.
Kimball, Sara E. 1994. The IE short diphthongs *oi, *ai, *ou and *au in Hittite. Sprache 36. 1-28.
Kimball, Sara E. 1999. Hittite historical phonology (Innsbrucker Beiträge zur Sprachwissenschaft
95). Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck.
Klingenschmitt, Gert. 1970. Griechisch ἱ ά κε θα . Münchener Studien zur Sprachwissenschaft 28.
75-88.
Klingenschmitt, Gert. 1982. Das altarmenische Verbum. Wiesbaden: Dr. Ludwig Reichert Verlag.
Kloekhorst, Alwin. 2008. Etymological dictionary of the Hittite inherited lexicon (Leiden Indo-
European Etymological Dictionary Series 5). Leiden: Brill.
Kluge/Seebold = Kluge, Friedrich. 2002. Etymologisches Wörterbuch der deutschen Sprache. 24.
durchgesehene und erw. Aufl., bearb. von Elmar Seebold. Berlin: Walter de Gruyter.
Kolb, Eduard. 1957. Alemannisch-nordgermanisches Wortgut (Beiträge zur schweizerdeutschen
Mundartforschung 6). Frauenfeld: Huber.
Kotin, Michael L. 2012. Gotisch. Im (diakronischen und typologischen) Vergleich (Sprache –
Literatur und Geschichte. Studien zur Linguistik/Germanistik 41). Heidelberg: Universitätsverlag
Winter.
Kortlandt, Frederik. 1979. Three problems of Balto-Slavic phonology. Zbornik za Filologiju i
Lingvistiku 22/2. 57-63.
Kortlandt, Frederik. 1990. The Germanic third class of weak verbs. NOWELE 15. 3-10.
Kortlandt, Frederik. 2008. Balto-Slavic phonological developments. Baltistica 43/1. 5-15.

199
Krahe, Hans. 1966. Germanische Sprachwissenschaft I (Sammlung Göschen 238). 6. Aufl. Berlin:
Walter de Gruyter.
Krahe, Hans. 1967. Germanische Sprachwissenschaft II (Sammlung Göschen 780). 6. Aufl. Berlin:
Walter de Gruyter.
Krahe, Hans & Wolfgang Meid. 1967. Germanische Sprachwissenschaft III (Sammlung Göschen
1218/1218a/1218b). Berlin: Walter de Gruyter.
Krause, Wolfgang. 1968. Handbuch des Gotischen. 3. neubearb. Aufl. München: C.H. Beck’sche
Verlagsbuchhandlung.
Kronasser, Heinz. 1962-1966. Etymologie der hethitischen Sprache. Wiesbaden: Otto Harrassowitz.
Kroonen, Guus. 2010. On Gothic iup and the Germanic directionals. NOWELE 58/59. 367-379.
Kroonen, Guus. 2011. Neuniederländisch zeug ‘Sau’ und spugen ‘speien’: Zwei Beispiele der
westgermanischen Velarisierung. In Guus Kroonen et al. (eds.), Thi limit lof. Festschrift für
Arend Quak zum 65. Geburtstag (Amsterdamer Beiträge zur Älteren Germanistik 67), 149-161.
Amsterdam & New York: Rodopi.
Kroonen, Guus. 2012. Non-Indo-European root nouns in Germanic: Evidence in support of the
Agricultural Substrate Hypothesis. In Riho Grünthal & Petri Kallio (eds.), A linguistic map of
prehistoric Northern Europe (Suomalais-Ugrilaisen Seuran Toimituksia / Mémoires de la Société
Finno-Ougrienne 266), 239-260. Helsinki: Société Finno-Ougrienne.
Kroonen, Guus. 2013. Etymological dictionary of Proto-Germanic (Leiden Indo-European
Etymological Dictionary Series 11). Leiden: Brill.
Kuiper, Franciscus B.J. 1942. Notes on Vedic noun-inflexion. Mededelingen der Koninklijke
Nederlandse Akademie van Wetenschappen, Afd. Letterkunde 5/4. 161-256.
Kuiper, Franciscus B.J. 1956. The etymology of ἄνθρωπος. In Heinz Kronasser (ed.), ΜΝΗΜΗΣ
ΧΑΡΙΝ. Gedenkschrift Paul Kretschmer 2. Mai 1886 – 9. März 1956 I. Wien: Verlag der Wiener
Sprachgesellschaft.
Kuryłowicz Jerzy. 1968. Indogermanische Grammatik 2. Heidelberg: Carl Winter.
Lambert, Pierre-Yves. 2003. La langue gauloise. Éd. rev. et augm. Paris: Errance.
Larsson, Jenny H. 2010. Proto-Indo-European root nouns in the Baltic languages. Copenhagen
University.
Lehmann, Winfred P. 1955. Proto-Indo-European phonology. Austin: University of Texas Press.
Lehmann, Winfred P. 1986. A Gothic etymological dictionary. Based on the 3rd ed. of
Vergleichendes Wörterbuch der gotischen Sprache by Sigmund Feist. Leiden: E.J. Brill.

200
Leumann, Manu et al. 1963. Lateinische Grammatik I (Handbuch der Altertumswissenschaft 2.
Abt., 2. Teil, 1. Bd.). Unveränderter Abdruck aus Stolz-Schmalz, Lateinische Grammatik, in 5.
Aufl. völlig neu bearb. von Manu Leumann und Joh. B. Hofmann 1926-1928. München: C.H.
Beck’sche Verlagsbuchhandlung.
Lindeman, Fredrik O. 1987. Introd ct on to t e ’Lar ngeal T eor ’. Oslo: Norwegian University
Press.
LIV = Rix, Helmut et al. 2001. LIV. Lexikon der indogermanischen Verben. Die Wurzeln und ihre
Primärstammbildungen. 2., erweiterte und verbesserte Aufl. bearb. von Martin Kümmel und
Helmut Rix. Wiesbaden: Dr. Ludwig Reichert Verlag.
Lloyd, Albert L. & Otto Springer. 1988. Etymologisches Wörterbuch des Althochdeutschen I. a –
bezzisto. Göttingen: Vandenhoeck & Ruprecht.
Lloyd, Albert L. et al. 1998. Etymologisches Wörterbuch des Althochdeutschen II. bî – ezzo.
Göttingen: Vandenhoeck & Ruprecht.
Lloyd, Albert L. & Rosemarie Lühr. 2007. Etymologisches Wörterbuch des Althochdeutschen III.
fadum – fûstslag. Göttingen: Vandenhoeck & Ruprecht.
Lühr, Rosemarie. 1976. Germanische Resonantengemination durch Laryngal. Münchener Studien
zur Sprachwissenschaft 35. 73-92.
Lühr, Rosemarie. 2000. Die Gedichte des Skalden Egill (Jenaer indogermanische Textbearbeitung
1). Dettelbach: J.H. Röll.
Marstrander, Carl. J.S. 1911. The deaths of Lugaid and Derbforgaill. Ériu 5. 201-218.
Matzel, Klaus. 1962. Anlautendes þl- und fl- im Gotischen. Die Sprache 8. 220-237.
Meillet, Antoine. 1898 Étymologies arméniennnes. Mémoires de la Société linguistique de Paris 10.
274-282.
Meillet, Antoine. 1922. Caractères généraux des langues germaniques. 2. éd., rev., corr. et augm.
Paris: Librairie Hachette.
McGuire, Brian P. 2008. Da Himmelen kom nærmere: fortællinger om Danmarks kristning 700-
1300. Frederiksberg: Alfa.
Neri, Sergio. 2003. I sostantivi in -u del gotico. Morfologia e preistoria (Innsbrucker Beiträge zur
Sprachwissenschaft 108). Innsbruck: Institut für Sprachen und Literaturen der Universität
Innsbruck.
Nielsen, Niels Åge. 2000. Dansk etymologisk ordbog. Ordenes historie. 4. udg., 5. opl. Kbh.:
Gyldendal.

201
Nielsen Whitehead, Benedicte. 2010. The ablaut of IE root nouns. Handout from the ‘Roots of
Europe evaluation meeting’ 14 October 2010 niversity of Copenhagen.
Nielsen Whitehead, Benedicte. 2013. Root-structure and ablaut in PIE root nouns. Handout from
‘The lengthened grade in Indo-European’ Leiden 29-31 July 2013.
Nikolaev, Alexander S. 2007. The name of Achilles. In Coulter H. George et al. (eds.), Greek and
Latin from an Indo-European perspective (Cambridge Classical Journal: Proceedings of the
Cambridge Philological Society, supplementary vol. 32), 162-173. Cambridge: The Cambridge
Philological Society.
NIL = Wodtko, Dagmar S. et al. 2008. Nomina im indogermanischen Lexikon. Universitätsverlag
Winter.
Noreen, Adolf. 1923. Altisländische und altnorwegische Grammatik (Laut- und Flexionslehre)
unter Berücksichtigung des Urnordischen (Sammlung kurzer Grammatiken germanischer
Dialekte 4: Altnordische Grammatik 1) 4., vollständig umgearb. Aufl. Halle (Saale): Verlag von
Max Niemeyer.
Nübling, Damaris. 2008. Was tun mit Flexionsklassen? Deklinationsklassen und ihr Wandel im
Deutschen und seinen Dialekten. Zeitschrift für Dialektologie und Linguistik 65. 282-330.
Nussbaum, Alan J. 1986. Head and horn in Indo-European (Studies in Indo-European Language
and Culture 2). Berlin: Walter de Gruyter.
Olander, Thomas. 2012. Proto-Indo-European *-os in Slavic. Russian Linguistics 36/3. 319-341.
Olander, Thomas. 2014. Proto-Slavic inflectional morphology. Copenhagen: Department of
Scandinavian Studies and Linguistics, University of Copenhagen.
Olsen, Birgit A. 1985. On the development of Indo-European prothetic vowels in Classical
Armenian. Revue des Études Arméniennes. Nouvelle Série 19. 5-17.
Olsen, Birgit A. 1988. The Proto-Indo-European instrument noun suffix *-tlom and its variants
(Historisk-filosofiske Meddelelser, Det Kongelige Danske Videnskabernes Selskab 55).
Copenhagen: Munksgaard.
Olsen, Birgit A. 1988-1989. A surrejoinder to J.A.C. Greppin’s remarks on prothetic vowels in
Armenian. Revue des Études Arméniennes 21. 481-483.
Olsen, Birgit A. 1994. The stages of IE aspiration by laryngeal. In George E. Dunkel et al. (eds.),
Früh-, Mittel-, Spätindogermanisch. Akten der IX Fachtagung der Indogermanischen
Gesellschaft vom 5. bis 9. Oktober 1992 in Zürich, 267-277. Wiesbaden: Dr. Ludwig Reichert
Verlag.

202
Olsen, Birgit A. 1999. The noun in Biblical Armenian. Origin and word formation (Trends in
Lingustics. Studies and Monographs 119). Berlin/New York: Mouton de Gruyter.
Olsen, Birgit A. 2008. On the origin and morphological ramification of the Greek type ἱππεύ .
Handout from the ‘XIII Fachtagung der Indogermanischen Gesellschaft’ 21-27 September 2008,
University of Salzburg.
Olsen, Birgit A. 2010. Derivation and composition. Two studies in Indo-European word formation
(Innsbrucker Beiträge zur Sprachwissenschaft 136). Innsbruck: Institut für Sprachen und
Literaturen der Universität Innsbruck.
Orel, Vladimir. 2003. A handbook of Germanic etymology. Leiden: Brill.
Otrębski Jan. 1965 ra at ka j zyka litewskiego II. Warszawa: Państwowe Wydawnictwo Nauk.
Pedersen, Holger. 1909. Vergleichende Grammatik der keltischen Sprachen I. Göttingen:
Vandenhoeck und Ruprecht.
Pedersen, Holger. 1913. Vergleichende Grammatik der keltischen Sprachen II. Göttingen:
Vandenhoeck und Ruprecht.
Peters, Martin. 1980. Untersuchungen zur Vertretung der indogermanischen Laryngale im
Griechischen (Philosophisch-historische Klasse der Österreichischen Akademie der
Wissenschaften, Sitzungsberichte 377 / Veröffentlichungen der Komission für Linguistik und
Kommunikationsforschung, Heft 8). Wien: Verlag der Österreichischen Akademie der
Wissenschaften.
Philippa, Marlies et al. 2003. Etymologisch woordenboek van het Nederlands. A-E. Amsterdam:
Amsterdam University Press.
Philippa, Marlies et al. 2005. Etymologisch woordenboek van het Nederlands. F-Ka. Amsterdam:
Amsterdam University Press.
Philippa, Marlies et al. 2007. Etymologisch woordenboek van het Nederlands. Ke-R. Amsterdam:
Amsterdam University Press.
Philippa, Marlies et al. 2009. Etymologisch woordenboek van het Nederlands. S-Z. Amsterdam:
Amsterdam University Press.
Pinault, Georges-Jean. 1982. A neglected phonetic law: the reduction of the Indo-European
laryngeals in internal syllable before yod. In A. Ahlqvist (ed.), Papers from the 5th International
Conference on Historical Linguistics, 265-272. Amsterdam: John Benjamins Publishing
Company.

203
Pinault, Georges-Jean. 1997. Terminologie du petit bétail en tocharien. Studia Etymologica
Cracoviensia 2. 175-218.
Polomé, Edgar C. 1980. Armenian and the Proto-Indo-European laryngeals. In John A.C. Greppin
(ed.), First International Conference on Armenian Linguistics: Proceedings. The University of
Pennsylvania, Philadelphia, 11-14 July 1979, 17-33. Delmar, NY: Caravan Books.
Praust, Karl. 2005. W at reek can tell s abo t t e re stor of ngl s ‘to fall’. Paper at the
conference ‘Greek and Latin from an Indo-European perspective’, Cambridge, July 2005.
Pronk, Tijmen. 2008. Sanskrit (v)r̥ṣabhá-, Greek ἄρ ην, ἔρ ην: the spraying bull of Indo-
European? Handout from ‘Indogermanistik und Linguistik im Dialog: XIII. Fachtagung der
Indogermanischen Gesellschaft’, 21.-27. September 2008, Salzburg.
Puhvel, Jaan. 1991 Hittite etymological dictionary 3. Berlin: Mouton de Gruyter.
Ramat, Paolo. 1981. Einführung in das Germanische (Linguistische Arbeiten 95). Tübingen: Max
Niemeyer.
Rasmussen, Jens E. 1978. Zur Morphophonemik des Urindogermanischen. Die Erklärung
qualitativer, quantitativer und akzentueller Alternationen durch vorurindogermanische
Lautgesetze. Collectanea Indoeuropaea 1. 59-139. [Reprinted in: Jens E. Rasmussen (ed.). 1999.
Selected papers on Indo-European linguistics. With a section on comparative Eskimo linguistics
(Copenhagen Studies in Indo-European 1), 1-66. Copenhagen: Museum of Tusculanum Press.]
Rasmussen, Jens E. 1983. Determining proto-phonetics: PIE laryngeals. In Fr. Karlsson (ed.),
Papers from the Seventh Scandinavian Conference of Linguistics, 371-384. Helsinki: University
of Helsinki, Department of General Linguistics. [Reprinted in: Jens E. Rasmussen (ed.). 1999.
Selected papers on Indo-European linguistics. With a section on comparative Eskimo linguistics
(Copenhagen Studies in Indo-European 1), 67-81. Copenhagen: Museum of Tusculanum Press.]
Rasmussen, Jens E. 1988. Indo-European Ablaut -i- ~ -e-/-o-. Arbejdspapirer udsendt af Institut for
Lingvistik, Københavns Universitet 7. 125-142. [Reprinted in: Jens E. Rasmussen (ed.). 1999.
Selected papers on Indo-European linguistics. With a section on comparative Eskimo linguistics
(Copenhagen Studies in Indo-European 1), 312-326. Copenhagen: Museum of Tusculanum
Press.]
Rasmussen, Jens E. 1989. Studien zur Morphophonemik der indogermanischen Grundsprache
(Innsbrucker Beiträge zur Sprachwissenschaft 55). Innsbruck: Institut für Sprachwissenschaft der
Universität Innsbruck.

204
Rasmussen, Jens E. 1992. One type of o-grade: a consonantal root infix? In Robert S.P Beekes et al.
(eds.), Rekonstruktion und relative Chronologie. Akten der VIII. Fachtagung der
Indogermanischen Gesellschaft. Leiden, 31. August - 4. September 1987 (Innsbrucker Beiträge
zur Sprachwissenschaft 65), 335-357. Innsbruck: Institut für Sprachwissenschaft der Universität
Innsbruck.
Rasmussen, Jens E. 1996. Szemer nyi’s theory on Indo-European i- and u-stems. Copenhagen
Working Papers in Linguistics 4. 137-141.
Rasmussen, Jens E. 2003. An integrated view on ablaut and accent in Indo-European. In Brigitte
L.M. Bauer & Georges-Jean Pinault (eds.), Language in time and space. A festschrift for Werner
Winter on the occasion of his 80th birthday (Trends in Linguistics. Studies and Monographs
144), 351-358. Berlin & New York: Mouton de Gruyter.
Ringe, Donald A. 1988. Laryngeal isoglosses in the Western Indo-European languages. In Alfred
Bammesberger (ed.), Die Laryngaltheorie und die Rekonstruktion des indogermanischen Laut-
und Formensystems, 415-441. Heidelberg: Carl Winter Universitätsverlag.
Risch, Ernst. 1974. Wortbildung der homerischen Sprache. 2., völlig überbearb. Aufl. Berlin: Walter
de Gruyter.
Rix, Helmut. 1970. Anlautender Laryngal vor Liquida oder Nasalis sonans im Griechischen.
Münchener Studien zur Sprachwissenschaft 27. 79-110.
Schaffner, Stefan. 2001. Das Vernersche Gesetz und der innerparadigmatische Wechsel des
Urgermanischen im Nominalbereich (Innsbrucker Beiträge zur Sprachwissenschaft 103).
Innsbruck: Institut für Sprachen und Literaturen der Universität Innsbruck.
Scheungraber, Corinna. 2012. Univerbation of prefixed verbs in the prehistory of Germanic: Goth.
fraisan ‘vers c en’ and c. *fāha-/fanga- ‘fangen’. Handout from ‘Etymology and the
European lexicon. 14th Fachtagung of the Indogermanische Gesellschaft‘ Copenhagen, 17.-22.
September 2012.
Schindler Jochem. 1972. L’apophonie des noms-racines indo-européens. Bulletin de la Société de
linguistique de Paris 67. 31-38.
Schindler, Jochem. 1972a. Das Wurzelnomen im Arischen und Griechischen. Würzburg: Inaugural-
Dissertation zur Erlangung der Doktorwürde der Philosophischen Fakultät der Julius-
Maximilians-Universität zu Würzburg.
Schindler, Jochem. 1975a. L’apophonie des thèmes indo-européens en -r/n. Bulletin de la Société de
Linguistique de Paris 70. 1-10.

205
Schindler, Jochem. 1975b. Zum Ablaut der neutralen s-Stämme des Indogermanischen. In Helmut
Rix (ed.), Flexion und Wortbildung. Akten der V. Fachtagung der Indogermanischen
Gesellschaft. Regensburg, 9.-14. September 1973, 259-267. Wiesbaden: Reichert Verlag.
Schindler, Jochem. 1976. On the Greek type ἱππεύ . In Anna M. Davies & Wolfgang Meid (eds.),
Studies in Greek, Italic and Indo-European linguistics offered to Leonard R. Palmer on the
occasion of his seventieth birthday, June 5, 1976 (Innsbrucker Beiträge zur Sprachwissenschaft
16), 349-352. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck.
Schrijver, Peter. 1991. The reflexes of the Proto-Indo-European laryngeals in Latin (Leiden Studies
in Indo-European 2). Amsterdam: Rodopi.
Schrijver, Peter. 1997. Some Western European substratum words. In Alexander Lubotsky (ed.),
Sound law and analogy. Papers in honor of Robert S.P. Beekes on the occasion of his 60th
birthday (Leiden Studies in Indo-European 9). Amsterdam: Rodopi.
Schwyzer, Eduard. 1959. Griechische Grammatik auf der Grundlage von Karl Brugmanns
griechischer Grammatik I (Handbuch der Altertumswissenschaft II, 1, 1). München: C.H.
Beck’sche Verlagsbuchhandlung.
Seebold, Elmar. 1967. Die Vertretung von idg. g h im Germanischen. Zeitschrift für vergleichende
Sprachforschung 81. 104-133.
Seebold, Elmar. 1970. Vergleichendes und etymologisches Wörterbuch der germanischen starken
Verben (Janua Linguarum, studia memoriae Nicolai van Wijk dedicata. Series practica 85). The
Hague: Mouton.
Seebold, Elmar. 1980. Materialsammlung. Die Vertretung von anlautend idg. g h und g im
Germanischen. In Manfred Mayrhofer et al. (eds.), Lautgeschichte und Etymologie. Akten der VI.
Fachtagung der Indogermanischen Gesellschaft. Wien, 24.-29. September 1978, 450-484.
Wiesbaden: Dr. Ludwig Reichert Verlag.
Sehrt, Edward H. 1966. Vollständiges Wörterbuch zum Heliand und zur altsächsischen Genesis. 2.
durchgesehene Aufl. Göttingen: Vandenhoeck & Ruprecht.
Sihler, Andrew L. 1995. New comparative grammar of Greek and Latin. New York: Oxford
University Press.
Solmsen, Felix. 1909. Beiträge zur griechischen Wortforschung I. Strassburg: Karl J. Trübner.
Specht, Franz. 1939. Sprachliches zur Urheimat der Indogermanen. Zeitschrift für vergleichende
Sprachforschung auf dem Gebiete der indogermanischen Sprachen 66. 1-74.

206
Stang, Christian S. 1966. Vergleichende Grammatik der Baltischen Sprachen. Oslo: Univer-
sitetsforlaget.
Stenton, Frank M. 1971. Anglo-Saxon England. 3. ed. Oxford: Clarendon Press.
Streitberg, Wilhelm A. 1896. Urgermanische Grammatik. Einführung in das vergleichende Studium
der altgermanischen Dialekte. Heidelberg: C. Winter.
Strid, Jan Paul. 2008. De origine gothorum. In Mette Bruus et al. (eds.), Gotisk Workshop. Et
uformelt formidlingstræf (Mindre skrifter udgivet af Center for Middelalderstudier, Syddansk
Universitet, Odense 26), 23-34. Odense: Center for Middelalderstudier, Syddansk Universitet,
Odense.
Sturtevant, Edgar H. 1947. An introduction to linguistic science. New Haven: Yale University Press.
Szemerényi, Oswald J.L. 1957. The Greek nouns in -εύ . In Mn s c ar n. edenksc r ft a l
Kretschmer, 2. Mai 1866-9. März 1956, Bd. II, 159-181. Wien: Wiener Sprachgesellschaft.
Szemerényi, Oswald. 1967. The history of Attic οὖς and some of its compounds. Studi Micinei ed
Egeo-Anatolici 3. 47-88.
Szemerényi, Oswald J.L. 1970. Einführung in die vergleichende Sprachwissenschaft. Darmstadt:
Wissenschaftliche Buchgesellschaft.
Thöny, Luzius. 2013. Flexionsklassenübertritte. Zum morphologischen Wandel in der
altgermanischen Substantivflexion (Innsbrucker Beiträge zur Sprachwissenschaft 146).
Innsbruck: Institut für Sprachen und Literaturen der Universität Innsbruck.
Thurneysen, Rudolf. 1975. A grammar of Old Irish. Rev. ed. Dublin: Dublin Institute for Advanced
Studies.
Tristram Hildegard. 2007. Why don’t the English speak Welsh? In Nick J. Higham (ed.) The
Britons in Anglo-Saxon England, 192-214. Rochester, NY: Boydell Press.
De Vaan, Michiel. 2003. The Avestan vowels (Leiden Studies in Indo-European 12). Amsterdam:
Rodopi.
De Vaan, Michiel. 2008. Etymological dictionary of Latin and the other Italic languages (Leiden
Indo-European Etymological Dictionary Series 7). Leiden: Brill.
Vaillant, André. 1950. Grammaire comparée des langues slaves I. Lyon: IAC.
Vaillant, André. 1958. Grammaire comparée des langues slaves II/1. Lyon: IAC.
Vaillant, André. 1974. Grammaire comparée des langues slaves IV. Lyon: IAC.

207
Vennemann, Theo. 1995. Etymologische Beziehungen im Alten Europa. In Jarmo Korhonen &
Jorma Koivulehto (eds.), Der Ginkgo Baum: germanistisches Jahrbuch für Nordeuropa 13.
Helsinki: Finn Lectura. [Reprinted in: Patrizia Noel Aziz Hanna (ed.). 2003. Europa Vasconica –
Europa Semitica, 203-297. Berlin.]
Vine, Brent. 1999. Greek ῥίζα ‘root’ and “schwa secundum”. UCLA Indo-European Studies 1. 1-27.
de Vries, Jan. 1962. Altnordisches etymologisches Wörterbuch. 2. verb. Aufl. Leiden: Brill.
Wackernagel, Jakob & Albert Debrunner. 1896. Altindische Grammatik I. Göttingen: Vandenhoeck
& Ruprecht.
Wackernagel, Jakob & Albert Debrunner. 1929. Altindische Grammatik III. Göttingen:
Vandenhoeck & Ruprecht.
Wackernagel, Jacob & Albert Debrunner. 1954. Altindische Grammatik II/2. Göttingen:
Vandenhoeck & Ruprecht.
Watkins, Calvert. 1969. Indogermanische Grammatik III/1. Heidelberg: Carl Winter.
Weitenberg, Joseph J.S. 1984. Die hethitischen U-Stämme (Amsterdamer Publikationen zur Sprache
und Literatur 52). Amsterdam: Rodopi.
Wessén, Elias. 1958. Isländsk grammatik. 2. uppl. Stockholm: Svenska Bokförlaget/Norstedts.
Widmer, Paul. 2004. Das Korn des weiten Feldes. Interne Derivation, Derivationskette und
Flexionsklassenhierarchie: Aspekte der nominalen Wortbildung im Urindogermanischen
(Innsbrucker Beiträge zur Sprachwissenschaft 111). Innsbruck: Institut für Sprachen und
Literaturen der Universität Innsbruck.
Van Windekens, Albert J. 1979. Le tokharien, confronté avec les autres langues indo-européennes
II/1 (Travaux publi s par le Centre International de Dialectologie G n rale de l’ niversit
catholique néerlandaise de Louvain 12). Louvain: Centre International de Dialectologie
Générale.
Winter, Werner. 1965. Armenian evidence. In Werner Winter (ed.), Evidence for laryngeals (Janua
Linguarum, studia memoriae Nicolai van Wijk dedicata. Series maior 11), 100-115. Mouton.
Zimmer, Stefan. 1981. Idg. *ukson-. Zeitschrift für vergleichende Sprachforschung 95. 84-91.

208
Abstract (English)

The present Ph.d. thesis entitled Archaisms and innovations: four interconnected studies on
Germanic historical phonology and morphology consists of two articles on archaisms and
innovations in the derivational system of Germanic nouns and two articles on regular sound changes
that may support the conclusions reached in the former two articles.
The first article Layers of root nouns in Germanic: chronology, structure and origin suggests that
we can recognise Germanic root nouns inherited from Proto-Indo-European by the circumstance
that their radical ablaut grade is predicted by the phonotaxis of their root. Germanic root nouns
violating these principles have entered the inflectional class of the root nouns at a later stage. Few
exceptions exist, but these may all be accounted for in various ways. The seeming exception of PG
*aik- f. ‘oak’ deserves special mentioning in that the new regular sound change of PIE *#h2i- > PG
*#ai-, which I suggest in the third article The outcome of PIE *#Hi- and *#Hu- in Germanic along
with a parallel sound change of PIE *#h2u > PG *#au-, allows for the interpretation of PG *aik- as
PIE *h2 ĝ- which is the shape expected for a root noun of that phonotactic structure.
The second article The structure, form and function of the Germanic primary i-stems also centres
on derivational morphology; in this case, however, with the i-stems as the object of study. For the
primary i-stems, I tentatively suggest that the archaic-looking types of i-stem adjectival agent nouns
and i-stem action nouns or verbal abstracts that occur with both radical o-grade and zero grade
originate from only one type in which, as was the case with the root nouns, the choice of radical
ablaut grade depends on the phonotaxis of the root. When handed down to Germanic, however this
system was abandoned and shaped anew in a manner compatible with the ablaut system of the
Germanic strong verbs, thereby giving rise to new radical ablaut grades in the primary i-stems.
In order to obtain a better understanding of possible transitions of nouns between i-stems and
other inflectional classes, the fourth article T e o tco e of I *- and *- n er an c
makes a few adjustments to the standard assumption on the history of some of the i-stem case
endings by suggesting a series of regular sound changes pertaining to PIE e-vowel diphthongs in
final syllables.
With its primary focus on how to identify archaisms and subsequently how to separate them
from later innovations in Germanic derivational morphology, this thesis will hopefully contribute to
the further development of the general field of Germanic and Indo-European studies by identifying
and removing some such forms that can easily be disregarded as innovations by future scholarship.

209
Abstract (Danish)

Denne ph.d.-afhandling med titlen Archaisms and innovations: four interconnected studies on
Germanic historical phonology and morphology består af to artikler om arkaismer og innovationer i
det germanske nominalderivationssystem og to artikler om nye lydlove, som kan tjene til støtte for
de konklusioner, der nås i de forudgående to artikler.
Den første artikel Layers of root nouns in Germanic: chronology, structure and origin lægger op
til, at vi kan genkende nedarvede germanske rodnominer på, at deres aflydstrin kan forudsiges ud
fra deres rodstruktur. Hvis et germansk rodnomen ikke lever op til de fastsatte kriterier, må vi kunne
forvente, at de er indtrådt i sproget på et senere tidspunkt, og at de således er innovationer. Dog er
der nogle få undtagelser, men disse kan der alle redegøres for ad diverse veje. Germ. *aik- f. ’eg’
som ved et første øjekast synes at udgøre en undtagelse, kræver en særlig kommentar, idet den nye
lydlov ie. *#h2i- > germ. *#ai-, som jeg foreslår i tredje artikel The outcome of PIE *#Hi- and
*#Hu- in Germanic sammen med den parallelle lydlov ie. *#h2u > germ. *#au-, åbner op for en
fortolkning af germ. *aik- som ie. *h2 ĝ-, hvilket bringer dette leksem i overensstemmelse med de i
den første artikel fastlagte kriterier.
Den anden artikel The structure, form and function of the Germanic primary i-stems handler
også om derivationsmorfologi; her dog med i-stammerne som primært forskningsobjekt. For i-
stammerne foreslår jeg, at de arkaisk udseende typer af adjektiviske nomina agentis og nomina
actionis eller verbalabstrakter, som foreligger med både o- og nultrin i roden, oprinder i én og
samme type, hvor rodstrukturen er afgørende for, hvilket aflydstrin i-stammen vil udvise. I
germansk bliver dette system opgivet og omarrangeret til at være afhængigt af aflydssystemet i de
stærke verber, som de knytter sig til. Herved opstår mulighed for nye aflydstrin blandt i-stammerne.
For at vi bedre skal kunne forstå de mulige stammeklasseovergange mellem i-stammerne og
andre fleksionsklasser, justerer den fjerde og sidste artikel The o tco e of I *- and *-
n er an c en anelse på nogle af de i forvejen kendte processer vedrørende et par i-
stammekasusendelser. Nærmere bestemt foreslår jeg en række nye lydlove, der opererer i e-
diftonger i final stavelse.
Med sit primære fokus på, hvordan arkaismer kan idenficeres og adskilles fra senere
innovationer vil denne afhandling forhåbentlig bidrage til germanistikkens og indoeuropæistikkens
videre udvikling ved at gøre opmærksom på sådanne former, som den fremtidige forskning let kan
se bort fra og således ikke længere behøver at bekymre sig om.

210