A Grammar of Proto-Germanic

Winfred P. Lehmann
Jonathan Slocum, ed.
III. INFLECTIONAL MORPHOLOGY
3.A.1. Introduction on Syntax
For understanding the syntax of Proto-Germanic it is highly important to note that Pre-IndoEuropean was an Active Language, and that the change to Accusative structure in Proto-IndoEuropean and its dialects was gradual. Moreover, Germanic separated early from the other IndoEuropean dialects, as I pointed out in 1953 and as others have since; more traits of Active Language
structure were maintained in it than in the dialects that had been assumed to be the most archaic,
that is, Indo-Iranian and Greek, on which Brugmann's Indo-European forms were based. While the
reconstructions of Brugmann and later Indo-Europeanists have been predominant in subsequent
handbooks of the parent language and also for explaining forms in the dialects, many of them do not
apply to Proto-Germanic.
Moreover, the features that Germanic shares with Celtic on the one hand, Italic on the other, and
with Balto-Slavic are chiefly lexical and relatively late, such as the adoption of the word for iron
from Celtic; these will be discussed in the chapter on semantics. It is also important to note that
there are no innovations common to these four dialects, a further indication that Germanic was
independent of them in its early phase. Proto-Germanic must accordingly be treated as a stage in the
development from Pre-Indo-European through Proto-Indo-European to distinct dialects.
Some characteristics of Active languages are stated here; others will be discussed in the chapters
dealing with derivational morphology and with sentence structure. Active languages have three
word classes: nouns, verbs, and particles. Nouns and verbs fall into two classes, either
active/animate or inactive/stative. Some examples of the earlier active structure have survived
throughout the Indo-European dialects in twofold designations for items that may be considered to
be either in action or at rest. Among these are the two words for fire. The active word, reflected by
Lat. ignis, Skt agns, Lith. ugns, OCS ogn, indicated actual burning; its active role is supported by
the use of the Sanskrit word also for the god of fire. The word reflected by Hitt. pahhuwar, Gk pr,
Umb. pir, Arm. hur, Toch. A por, as well as by the Germanic words as maintained by NE fire,
indicated a state, as supported by its neuter gender. Other examples will be given in the chapter on
semantics. When the dialects shifted from active to accusative structure, the reason for maintaining
two words was lost, and typically only one was maintained.
As another characteristic of Active languages, sentences are constructed for the most part by pairing
active nouns with active verbs and stative nouns with stative verbs, that is, through agreement rather
than government. Accordingly, transitivity is not a characteristic and there is no inflection as for an
accusative case. Relationships among nouns, also with verbs, are indicated by means of particles;
furthermore, there is no distinct "adposition" part of speech.

In the course of time, agreement structure was gradually replaced by government structure in ProtoIndo-European and its dialects. Explanations for such changes are speculative; in my view, the
linguistic change resulted from cultural change: as civilization developed, speakers began to
understand causal relationships. For example, they began to account for increase in productivity of
crops through proper planting, watering, and application of fertilizer, rather than through prayer to a
crop deity. To represent such causal effects in the language, specific particles came to be joined to
nouns and to verbs, some of which indicated transitivity; these and others provided the bases of
inflection. Through accentuation, such groups of base words and particles became units. In time,
loss of unaccented syllables and other changes clouded earlier structures, obscuring their origins,
but residues are still evident in the dialects as noted below. By means of such events, the shift to
government structure can be more readily understood.
Some of the particles remained independent, maintaining their function of indicating relationships
of nouns and of verbs, while others were affixed to these (cf. Brugmann, Gdr. 1.3/2: 964-1009,
Lehmann 2002: 87-99). Examples of reflexes in the Germanic languages that remained independent
are: Go. ak, OE ac 'but'; OE and OHG nu 'now'. Among examples that were suffixed are: Go. -ei, a
particle suffixed to demonstrative pronouns to produce relative pronouns as in the accusative
singular an-ei 'whom, which'; Go. -u and -uh, a question particle that is placed after words, as in:
Mark 10:38
maguts-u driggkan stikl anei ik driggka?
can-you-two ? drink
cup which I drink
"Can you two drink the cup which I drink?"
Other particles were adapted as prepositions, such as Go. af and OE of 'of', Go. bi, OE and OHG b
'at, around', OHG zi 'to', which Hirt equated with the suffix in Gk okon-de 'homeward' (Hirt 1932,
II: 103).
In addition to the particles that have been maintained as independent forms, some have survived in
suffixes and endings. It is assumed that the earliest such suffixes added were the single consonants
called determinatives. Because they were added to roots in early Proto-Indo-European, their
meanings are for the most part no longer distinguishable. Among examples are forms made from the
Proto-Indo-European root *wer- 'turn, bend', as in the nouns Lat. vermis, Go. warms 'worm' and
possibly the Lith. vrti 'to bend' (Pokorny 1959:1152-1160). Among forms derived from the root
*wer- with determinatives are: Gk rmb 'turn about in a circle,' Go. wairpan, ON verpa, OE
weorpan, OHG werfen 'throw < turn' with PIE -b-determinative; Skt vartati 'turns', Go. wairan,
OHG werden 'become' with -t- determinative. Many such verbs made with determinatives could be
cited, e.g. Go. giutan, ON gjta, OHG giozzan 'pour', made with the -t- determinative from the root
PIE gew- as in Gk kh 'I pour'. But a semantic basis for its addition in the two roots and others
cannot be identified.
Some particles developed into suffixes consisting of syllables that have survived with specific uses
or meanings, such as the -to- and -no- suffixes on participles. Subsequently, many suffixes were
based on full words, for example the widely used NE -hood, Du. -heid, NHG -heit suffix that
indicates state or condition, as in childhood. The noun also remained independent in the older
Germanic dialects, as in Go. hidus 'manner', OHG heit 'rank, state, condition'.
Besides their adaptations in derivational morphology, particles were also the basis of inflectional

endings, such as -s, the nominative singular ending as in Go. dags 'day'. In his discussion of such
developments, Hirt pointed out that Bopp had already equated the -s ending with the Proto-IndoEuropean particle *so 'here', with reflexes in Skt sa, Go. sa, Gk ho; *so also became the basis of
pronouns (Hirt 1932, II:9; cf. also Hirt 1927, 3:13-14; Lehmann 2002: 148-149). Hirt assumed that
*so was the source of another case ending as well, that of the genitive singular as in Go. brors 'of
the brother'. Then at the conclusion of his treatment he proposed that it was also the source of a
derivational s- suffix, as maintained in NHG Fuchs 'fox', but he did not explain the basis of its
development. We assume that, in the active stage of Pre-Indo-European and early Proto-IndoEuropean, *so was placed after nouns to indicate the subject and also the agent of reference in
genitive constructions; when Proto-Indo-European became an Accusative language, it was attached
as an ending, with eventual loss of the -o.
Comparable to the -s ending that came to indicate relationships for nouns, the -m ending that came
to indicate first person in verbs has been related to the personal pronoun m- forms. Such
combinations became fixed, so that nouns and verbs consisted of roots and endings. Some nominal
residues from this period consisting only of root and ending survive among common words; they
are referred to as root nouns (Brugmann 1906, II.1:130-146). PGmc fts, Gk ps, Lat. ps, Skt pad< PIE ped- is among the few nominal examples; PGmc ist, Lat. est < PIE ?es-t- is a verbal example.
But most nouns and verbs have an additional suffix between the root or base and the ending, among
them even reflexes of root nouns. While a reflex of PGmc fts is maintained in OE ft 'foot' without
the subject marker, the Gothic cognate was an expanded u-stem noun, fotus. A reason for adding
such suffixes may be illustrated by noting Latin verb forms: the common verb fer- < PIE ber- has
as second and third singular forms fers, fert. If these endings were added directly to most Latin verb
roots, e.g. laudare in the first conjugation and habre in the second, the forms would be *lauds,
*laudt and *habs, *habt, or more likely modified forms of them in view of the final consonant
combinations. The inclusion of the suffixes -/a- and -/e- in these conjugations between the root
and the endings, as in Lat. lauds, laudat 'praises', habs, habet 'has', yielded transparent stems and
endings.
The most frequent such suffix added to nouns and to verbs, PIE -e/o-, is not associated with any
meaning. Its wide application may be understood when forms like those cited above are compared;
endings may no longer be modified, as are those added directly, such as Lat. es, Gk es < PIE es-si
'thou art'. Already in Proto-Indo-European, most nouns and verbs included such affixes when
inflected, whether simple vowels like -e/o-, -i-, -u-, or longer suffixes like -yo-, -wo-, -no-. Nominal
and verbal inflections were labeled by them on the basis of the Indo-European form of the suffix,
e.g. o-stems, yo-stems, etc. Eventually many of them became indistinct, especially in dialects that,
like Germanic, introduced a strong stress accent on roots which led to weakening or loss of final
syllables.
3.A.2. Inflectional Morphology; Classes of Words
The Proto-Germanic lexicon consists of two classes of inflected words and a number of uninflected
classes. The two inflected classes are substantives and verbs. The uninflected classes are
conjunctions, adverbs, interjections, and prepositions (earlier, postpositions).
Substantives, including nouns, pronouns, adjectives and numerals, are inflected primarily for case,

secondarily for gender and number. The sub-class of nouns is inflected for case, gender and number.
The sub-class of pronouns is inflected for case, but only defectively for number and gender as well
as person. The sub-class of adjectives is inflected for gender, as well as for case and number; it is
further distinguished by addition of suffixes to indicate comparison. Cardinal numerals have
defective inflection in all three categories. Ordinal numerals are inflected like adjectives, e.g. Go.
ridja as n-stem for the numeral 'third'.
Verbs are inflected for person and number, tense, mood and voice.
3.1. Inflection of Substantives.
Five cases are reconstructed for Proto-Germanic, with traces of a sixth; these are: nominative,
vocative, genitive, dative, accusative, and instrumental. The nominative is the case used to indicate
the subject, and never follows a preposition. The vocative is the case of address. The genitive
indicates relationships among substantives, often possession. In addition to being governed by
specific prepositions, the dative indicates the indirect object; the accusative, the direct object. The
instrumental case has a distinct form in only one paradigm; it indicates a relationship involving
means, similar to that of adverbs. Two further cases are reconstructed for Proto-Indo-European: the
locative, which according to some specialists has left reflexes in certain Germanic paradigms, and
the ablative, to which certain Germanic adverbs have been related.
There are three genders: masculine, feminine, and neuter. These categories are primarily
grammatical, although there is also relationship to sex. That distinction is found largely with nouns
referring to animate beings, as in Go. nijis 'male relative', nij 'female relative'; frijnds 'friend',
frijndi 'woman friend'; ON Freyr 'the god Frey', Freyja 'the goddess'. Gender distinction is also
found in the third person pronouns.
Substantives are inflected for three numbers: singular, plural, and dual. The dual is strongly
represented only in pronouns and in PGmc was losing ground there. Like gender, number is also
primarily a grammatical category, not always a category with literal meaning, as the following
examples from Old Saxon indicate:

OS to godes hsun (pl.) 'to the house of God'

OS uuas im helpono (pl.) tharf 'he needed help'
OS briost (pl.) 'breast'
OS giscapu (pl.) 'fate'

With the exception of gender, the categories of inflection were less distinct in Proto-Germanic than
in Proto-Indo-European and were reduced further in the dialects, where some of them were
ultimately lost, such as gender in English with covert usage persisting primarily in the use of
personal pronoun 'he, she, it'. Similarly, except in the personal pronouns only two cases remain in
English, the genitive and the unmarked case. Further, the category of number is overt only in the
noun, in a few verb forms like am, is in contrast with are, and in the indication of person in the third
singular present, e.g. writes as opposed to write in the plural and other categories.
Inflection is indicated through the suffixes known as endings. As noted above, in Proto-IndoEuropean and early Proto-Germanic the endings were suffixed directly to roots. But affixes were
added to roots already in Proto-Indo-European to form bases, also known as stems, and the endings

were attached to these. When the stress accent was introduced, it generally fell on the root; weakly
stressed syllables then were often reduced, so that the endings in Proto-Germanic and its dialects
consisted of merged suffixes and the early endings. Classes of inflections in late Proto-Germanic
were labeled by these. Three declensions then resulted:
1. Root nouns (which are poorly attested);
2. Consonant stems;
3. Vocalic stems in two sets: the o/ stems, and the vocalic resonant stems.
Reconstructed forms illustrating each of the noun classes are given in paradigms below. Reflexes in
the dialects are included, to provide evidence for the Proto-Germanic forms that have been
reconstructed. In the paradigms of the dialects attested forms are preferred. But our limited texts do
not provide us with complete inflections for most nouns; accordingly some unattested forms are
included without being starred, such as the Gothic nominative dags.
3.2. Inflection of Nouns
3.2.1. The Root Nouns

By the time of the Germanic dialects some of the endings had been lost, as in the nominative and
accusative Old English forms of 'foot'. Others were replaced, such as that in the Gothic nominative
singular fotus by u-stem endings. Old English and Old Norse have maintained enough of the
inflection in some words, such as the word for 'foot', to reconstruct it for Proto-Germanic. In Old
Norse, however, the genitive singular is based on that of the u-stems; in Old English it is based on
that of the o-stems.
The endings of root nouns:
Sing.
PIE
PGmc
Nom.
-s
-s
Gen.
-es
-ez
Dat.
-ey/-y
-i
Acc.
-m
-un
The forms of the word for foot:

Plur.
Nom.
Gen.
Dat.
Acc.

PIE
-es
-m
-mis
-ns

PGmc
-ez
-n
-miz
-unz

Sing.
PGmc
ON
OE
Plur.
PGmc
ON
OE
Nom.
fts
ftr
ft
Nom.
ftez
ftr
ft
Gen.
ftez
ftar
ftes
Gen.
ftn
fta
fta
Dat.
fti
fte
ft
Dat.
ftmiz
ftum
ftum
Acc.
ftun
ft
ft
Acc.
ftunz
ftr
ft
The inflection of the noun for fire, an Indo-European r/n-stem, is of interest in showing that the
paradigm was maintained in early Proto-Germanic, but then modified by analogy. Only the
nominative/accusative singular forms are given for Old Norse and the West Germanic languages
because the two other cases have adopted forms from the o-stems. In the Gothic nominative
singular the -n has been extended by analogy with the genitive and dative.
Sing.
Nom/Acc.
Gen.

PGmc
fr/fuir
funis

Go.
fn
funins

ON
frr, fyrr, funi

OE
fyr

OHG
fiur

Dat.

funi

funin

3.2.2. The Consonant Stems

In the consonant stems a suffix ending in a consonant was added to the root. The most prominent of
these suffixes was -en. This varied from -en- to -n- to -on- in accordance with the accent. Moreover,
some of the nouns to which it was added ended in a vowel, so that the variants -n- and -n- arose;
if these stood finally in a word, the -n was lost, leading to a great variety of endings. Such
phonological variation in a paradigm is open to analogical modification so that it is difficult to
reconstruct the original paradigm. According to Meillet this is best reflected in Gothic as illustrated
in its forms for 'ox', most of which are reconstructed (1937:301ff.). To illustrate the regularization in
the dialects, examples of forms of the Greek words for 'male' and 'stone' are given here; one vowel
has been generalized in these.
Nom.sg.
Dat.sg.
Nom.pl.
Gen.pl.

PIE ending
-n
-en-i
-on-es
-n-

Gothic 'ox'
auhsa*
auhsin*
auhsans*
auhsne

Greek 'male'
rsn
rseni
rsenes
arsnn

Greek 'stone'
kmn
kmoni
kmones
akmnn

3.2.3. The n- stems

To illustrate the inflection for masculine nouns a complete set of forms of the word for man is given
here with the endings reconstructed for Proto-Indo-European. The Nom/Acc. forms for the neuter
nouns are given below; their genitive and dative forms are like those of the masculine nouns.
The forms show instances of analogical regularization, as discussed in the notes.
Masculine
Nom.sg.
Gen.sg.
Dat.sg.
Acc.sg.
Nom.pl.
Gen.pl.
Dat.pl.
Acc.pl.

PIE
-n,-
-en-os
-en-i
-on-m
-on-es
-n-m
(-on-mis)
(-ns)

PGmc
gum
gumenaz
gumeni
gumanun
gumaniz
gumann
gumanmiz
gumanunz

Go.
guma
gumins
gumin
guman
gumans
guman
gumam
gumans

ON
gume
guma
guma
guma
gumar
guma
gumom
guma

OE
guma
guman
guman
guman
guman
gumena
gumum
guman

OHG
gomo
gomen
gomon
gomon
gomon
gomno
gomm
gomon

Neuter
PIE
PGmc
Go.
ON
OE
OHG
Nom/Acc.sg.
-
aug
ug
auga
age
ouga
Nom/Acc.pl.
-n-
augn
ugna
augo
agan
ougon
Except for the genitive plural ending -, the Gothic forms support those reconstructed for ProtoGermanic. (It may be useful to note here again that e and o in Gothic indicate long vowels so that a
macron is superfluous, though often added in this grammar to be explicit; the Gothic short vowel
counterparts are indicated by ai and au, often distinguished from the i and u diphthongs by an
acute accent on the second member, e.g. a a.)
Many of the forms in the other dialects show the results of analogical extension. In the singular the
Old Norse and Old English genitive and dative forms were taken over from the accusative. In the

nominative/accusative plural the Old High German -o- of the ending is generalized from the
singular. The Old Norse nominative plural ending is based on that of the o-stems, as is also the
accusative plural ending. In the other dialects the form of the nominative was generalized to the
accusative. In the genitive plural the Old English and Old High German forms were taken over from
the feminine, as was the dative plural in Old High German.
The Proto-Germanic forms of the feminine are reconstructed largely on the basis of those in Gothic.
In the other dialects the forms influenced one another; some were introduced from other inflections.
Feminine
PGmc
Go.
ON
OE
OHG
Nom.sg.
tungn
tugg
tunga
tunga
zunga
Gen.sg.
tungn(i)z
tuggns
tungo
tungan
zungn
Dat.sg.
tungni
tuggn
tungo
tungan
zungn
Acc.sg.
tungnun
tuggn
tungo
tungan
zungn
Nom.pl.
tungniz
tuggns
tungor
tungan
zungn
Gen.pl.
tungnom
tuggn
tungna
tungena
zungno
Dat.pl.
tungomm(iz)
tuggm
tungom
tungum
zungm
Acc.pl.
tungnz
tuggns
tungor
tungan
zungn
In the oblique cases the endings of the masculine may be compared with those of the feminine;
except for the characteristic long stem vowel in the feminines, the two inflections are identical. In
Old Norse the influence of the two inflections is reciprocal. Various explanations have been offered
for the - vowel in the Old High German forms; by one of them it was introduced from the
accusative singular in which the vowel resulted from u-modification, so that the Proto-Germanic
accusative singular ending developed from -num to -num > -nu > -n. It then was taken over for
the other cases in the singular and for the accusative plural.
3.2.4. The r- stems

In Proto-Indo-European the form of the stem in the r-stems varied considerably. With the strong
tendency to regularization in Germanic, the variation gradually was eliminated; but regularization
apparently was incomplete in Proto-Germanic: from the differing paradigms in the individual
dialects, it seems that the process must have continued after the time of Proto-Germanic.
Explanation of the individual forms is therefore problematic in the treatment of the dialects. The
word atta has replaced the inherited word for father in Gothic except for one instance of the
vocative form, fadar, so the word for brother is given here.
PGmc
Go.
ON
OE
OHG
Nom.sg.
far
brar
faer
fder
fater
Gen.sg.
farez
brrs
fur
fder(es)
fater(es)
Dat.sg.
faer(i)
brr
fer
fder
fater(e)
Acc.sg.
faerun
brar
fur
fder
fater
Nom.pl.
fa(e)riz
brrjus
fer
fderas
fatera
Gen.pl.
farn
brr
fera
fdera
fatero
Dat.pl.
farumis
brrum
ferum
fderum
faterum
Acc.pl.
faruns
brruns
fer
fderas
fatera
In the Gothic plural the nominative has adopted the u-stem ending, as have the dative and
accusative, with -u- in the dative probably on the pattern of the accusative. In Old Norse the base

vowel of the nominative plural was extended to the three other forms. The Old English and Old
High German forms illustrate how endings of the o-stems were extended to other paradigms.
3.2.5. The nt- stems

A small number of nouns with the same suffix as that of the present participle preserved their
autonomy in Proto-Germanic. The changes in base vowel in Old English provide evidence for the
earlier inflection. In ON frndi 'relative' the back vowel of the nominative-accusative plural has
been generalized to the other cases. The singular endings have been replaced by forms of the ninflection; in Old English and Old High German the endings have been taken over from the o-stems
as have the Gothic nominative and genitive singular, as well as the genitive and dative plural.
Nom.sg.
Gen.sg.
Dat.sg.
Acc.sg.
Nom.pl.
Gen.pl.
Dat.pl.
Acc.pl.

PGmc
frijund
frijundiza
frijundi
frijundun
frijundiz
frijundn
frijundmiz
frijundunz

Go.
frijnds
frijndis
frijnd
frijnd
frijnds
frijnd
frijndam
frijnds

ON
frndi
frnda
frnda
frnda
frndr
frnda
frndom
frndr

OE
frond
frondes
frend
frond
frend
fronda
frondum
frend

OHG
friunt
friuntes
friunte
friunt
friunt
friunto
friuntum
friunt

3.2.6. The s- stems

The s-inflection has been maintained even less distinctly than have the other consonantal
declensions; it is chiefly of interest as a source of the prominent noun plural ending -er in Old High
German and in the later language. While the stem suffix is attested in a few oblique singular forms
in Old High German, one of them chalbire 'to the calf', the singular endings were in general
replaced by those of the e/o-inflection, as also in Old English.
Nom/Acc.sg.
Gen.sg.
Dat.sg.
Nom/Acc.pl.
Gen.pl.
Dat.pl.

PGmc
lambaz
lambezaz
lambizi
lambaz
lambizn
lambizumiz

OE
lamb
lambes
lambe
lombru
lombra
lombrum

OHG
lamb
lambes
lambe
lembir
lembiro
lembirum

3.3. The Vowel Stems

In the vowel stems the endings were attached to specific vowels or to suffixes ending in vowels. In
these declensions the forms developed distinctive endings. The resultant forms became the most
prominent in the Germanic dialects. In masculine and neuter nouns the vowel was e/o; in the
feminine nouns the vowel was in Proto-Indo-European. These have come to be used in the labels
for the declensions concerned. Besides being suffixed to the roots of nouns, these vowels were also
combined with resonants, to yield the suffixes -ye/o-, -y- and -we/o-, -w-. These came to have
distinctive inflections in keeping with the allophonic variation of the resonants, e.g. [i y iy].
3.3.1. The o- stems

Masculine

PGmc

Go.

ON

OE

OHG

Nom.sg.
Gen.sg.
Dat.sg.
Acc.sg.
Nom.pl.
Gen.pl.
Dat.pl.
Acc.pl.

dagaz
dagaza
dagai
dagan
dagzez
dagm
dagamaz
daganz

dags
dagis
daga
dag
dags
dag
dagam
dagans

dagr
dags
dege
dag
dagar
daga
dgom
daga

dag
dages
dge
dag
dagas
daga
dagum
dagas

tag
tages
tage
tag
taga
tago
tagum
taga

Neuter
PGmc
Go.
ON
OE
OHG
Nom/Acc.sg.
wordan
ward
or
word
wort
Nom/Acc.pl.
word
warda
or
word
wort
The full reflex of the Proto-Germanic nominative singular ending is found in Runic inscriptions,
e.g. Krogstad stainaz 'stone', and of the accusative singular, as in the form staina from the Tune
inscription.
The genitive singular ending of the o-stems was -sa. It was added to both stem-vowels -a- and -e-.
The -asa form has reflexes in the general North Germanic form, as in the Valsfjorden example
godagas, and also in Old Saxon, e.g. dagas. The -esa form has reflexes in Gothic, and in the other
West Germanic forms.
The dative form has the ending -i; PGmc -ai > -a in Gothic and -e in the other dialects. The
accusative form underwent regular development.
In Old High German and Old Saxon some forms in -u are attested that are considered relic forms of
the instrumental; the ending is derived from PIE -.
The nominative plural forms in Gothic, Old Norse, and Old High German are regular reflexes of the
Proto-Germanic form. The Old English and Old Saxon forms on the other hand must have been
derived from a longer form, such as PGmc -zez, which is parallel to the Vedic extended plural
-sas. By another explanation, the Germanic form may have arisen independently as a double plural
like that in children.
The genitive plural forms reflect PIE -o-om > -m except for that of Gothic with -m. Numerous
explanations have been proposed for -m, as from other PIE forms than genitives, but since the form
occurs in Gothic alone it must have an explanation from patterning there. I assume it was a
spontaneous change in which the front vowel was introduced in contrast with the back vowel of
the nominative (Lehmann 1968).
The dative plural forms must be derived from the Proto-Indo-European instrumental ending -mis in
view of the forms of the personal names Vatvims and Aflims recorded in inscriptions and Runic, e.g.
Stentoften dat. pl. hagestumz 'stallions'.
The accusative plural forms in Gothic, Old Norse, and Old High German are direct reflexes of PIE
-ons; the Old English form is based on the nominative.
The nominative/accusative neuter has the same origin as the accusative singular masculine; the -a is
preserved in Gallehus horna. Similarly, the nominative/accusative plural has the same origin as the

nominative singular feminine, i.e. PIE - > PGmc -. In North Germanic it developed first to -u, as
indicated in the form brn < *barnu; it was also maintained in the West Germanic dialects after
short syllables, as in OE hofu. The genitive and dative neuter forms are like those of masculine
nouns. The neuter genitive and dative forms are like those of masculine nouns.
3.3.2. The - stems

The suffix of the -stems developed in early Indo-European from -e/o- followed by a laryngeal.
Initially, then, their inflection was like that of the e/o-stems, since the laryngeal consonant was
comparable to others like s. As the laryngeal that modified vowels, it changed the preceding vowel
to a, e.g. ah > . The combination of the contracted with the endings led to considerable
differences, as the following illustrate.
PGmc
Go.
ON
OE
OHG
Nom.sg.
geb
giba
gjf
giefu
geba
Gen.sg.
gebz
gibs
gjafar
giefe
geb
Dat.sg.
geby
gibi
gjf
giefe
gebu
Acc.sg.
gebn
giba
gjf
giefe
geba
Nom.pl.
gebz
gibs
gjafar
giefa
geb
Gen.pl.
gebn
gib
gjafa
giefa
gebno
Dat.pl.
gebmiz
gibm
gjfom
giefum
gebm
Acc.pl.
gebz
gibs
gjafar
giefa
geb
As these examples may indicate, most of the forms of this declension underwent regular changes in
the dialects, though some forms within the declension were replaced, such as the Old Norse and Old
English dative singular; the Old High German genitive singular was replaced by analogy with the
Old High German accusative. In West Germanic, -u was lost after long syllables, as in OE r
'honor'. The expected ending -u was preserved in the Opedal Runic form liubu 'dear'.
Other forms were replaced by those of another inflection. The Old Norse inflection introduced istem forms in the nominative and accusative plural gjafar. The Old High German genitive plural
was taken over from the n-stems.
3.3.3. The yo- and y- stems

The two suffixes were frequent in Proto-Indo-European and Proto-Germanic and did not lead to
separate declensions until the allophonic system of the resonants broke down and the resultant
phonemes underwent separate developments. Distinct declensions resulted, not only for the nouns
ending in these suffixes as opposed to those ending in -o/-, but also for nouns with short as
opposed to long stem syllables; /y/ would have had the allophone [y] after short stem syllables, [iy]
after long. The paradigms illustrate the developments.
Masculine nouns with long stem vowels and -iyo-; PGmc herdiyas 'shepherd'

Long Stem
Nom.sg.
Gen.sg.
Dat.sg.
Acc.sg.
Nom.pl.

PGmc
herdiyas
herdiyeza
herdiyai
herdiyam
herdiyzez

Go.
hardeis
hardeis
hardja
hardi
hardjs

ON
hirer
hires
hire
hire
hirar

OE
hierde
hierdes
hierde
hierde
hierdas

OHG
hirti
hirtes
hirte
hirti
hirte/-

Gen.pl.
Dat.pl.
Acc.pl.

herdiym
herdiyamiz
herdiyanz

hardj
hardjam
hardjans

hira
hirom
hira

hierda
hierdum
hierdas

hirt(e)o
hirtum
hirte/-a

Masculine nouns with short stem vowels and -yo-; PGmc heryaz 'army'

Short Stem
PGmc
Go.
ON
OE
OHG
Nom.sg.
heryaz
harjis
herr
here
heri (Neut.)
Gen.sg.
haryeza
harjis
herz
herijes
heries
Dat.sg.
haryai
harja
her
herije
herie
Acc.sg.
haryam
hari
her
here
heri
Nom.pl.
haryoz
harjos
hariar
herijas
heris
Gen.pl.
haryom
harje
heria
herija
herio
Dat.pl.
haryamiz
harjam
heriom
herijum
herium
Acc.pl.
haryanz
harjans
haria
herijas
heris
The difference between nouns with long and short stems is most clearly evident in the genitive
singular forms. Upon the loss of the final vowel, the -e- before -s became -i-, yielding -iyis, -yis. In
-iyis the -y- was lost, and the -ii- fell together with --, as indicated by Gothic -ei-. In the short
stems, -yis was maintained.
These inflections are also of interest in indicating the effects of analogy as sound changes disrupt
the original alignments. In Gothic the nominative singular form harjis was remodeled on the form
of the genitive after the pattern of the long stems. In Old High German the nominative singular form
was probably taken over from the accusative. By the ninth century the -j- was lost; except for the
mutated stem vowel and the nominative singular, the forms agreed with those of the o-stems.
3.3.4. The wo- and w- stems

By the time of the dialects, these stems had merged with the -o/-stems. Old Norse stem vowels,
however, illustrate the modifying effect of the suffix consonant, e.g. hrr < PGmc harwas 'flax'.
Although subsequent sound changes and analogical modifications have given rise to complex
paradigms, the Proto-Germanic inflections can be reconstructed with allophonic variations of -wparallel to those of -y-. Proto-Germanic forms for harwas are given here; after long bases, the suffix
would have been -uw-.
PGmc
Nom.
Gen.
Dat.
Acc.

Singular
harwaz
harweza
harwai
harwam

Plural
harwz
harwm
harwamiz
harwanz

3.3.5. The i- and u- stems

The -i- and -u- stems are comparable to the -o- stems, with -i- and -u- occupying the place of -o-.
Their combinations with other vowels, however, led to differences. If the principal accent came to
stand on the stem-vowel, the syllables would have had the shape -ey- in Proto-Indo-European; in
this way the nominative plural differed from that of the o-stems, and is reconstructed as -ey-es. The
suffix could also have o-grade, as in the feminine genitive singular, reconstructed as -oy-so-. The
dative singular was taken over from the o-stems in Gothic.

i-stems
PGmc
Go.
ON
OE
OHG
Nom.sg.
gastis/z
gasts
gestr
gst
gast
Gen.sg.
gastez
gastis
gests
gstes
gastes
Dat.sg.
gastai
gasta
gest
gste
gaste
Acc.sg.
gastin
gast
gest
gst
gast
Nom.pl.
gastiyiz
gasteis
gester
gstas
gesti
Gen.pl.
gastiyn
gast
gesta
gsta
gestio
Dat.pl.
gastimz
gastim
gestom
gstum
gestim
Acc.pl.
gastinz
gastins
geste
gstas
gesti
The full reflex of the Proto-Germanic nominative ending is found only in Runic inscriptions, e.g.
Gallehus hlewagastiz. In Old High German the stem vowel was lost after long stems before
modification of the base vowel, so that the singular does not have umlaut. The feminine survives in
a few forms, but in general it has merged either with the masculine or with other declensions.
u-stems
PGmc
Go.
ON
OE
OHG
Nom.sg.
sunuz
sunus
sunu
sunu
sun(u)
Gen.sg.
sunauz
sunus
suna
suna
sunes
Dat.sg.
sunu
sunu
suna
suna
sune
Acc.sg.
sunun
sunu
sun
sunu
sun(u)
Inst.sg.
suniway
suniu
Nom.pl.
suniwiz
sunjus
suna
suna
suni
Gen.pl.
suniwn
suniw
suna
suna
suneo
Dat.pl.
sunumiz
sunum
sunum
sunum
sunim
Acc.pl.
sununz
sununs
suna
suna
suni
In the Tomstad Runic inscription the full ending of the nominative singular was maintained in
waruz 'protection', and the accusative singular in the Kjlevig inscription form magu 'kinsman'. The
singular endings are as expected, though the Old English dative is taken over from the genitive. The
vocative singular, which survives in Gothic, has the same form as the accusative singular. The plural
forms are also as expected, though with influences among them; the Old High German nominative
has been taken over in the accusative, which, however, is still reflected in the form situ 'custom' in
Otfrid.
Forms of neuter u-stems are found only in the singular in Gothic, e.g. nominative fahu, genitive
fahus, dative fahu 'cattle'; cf. also filu 'much'. In the other dialects they have fallen together with
other declensions, or survive in relic forms.
3.3.6. Development of Noun Inflection in Proto-Germanic

The forms given above indicate the general course of development of noun inflection in ProtoGermanic: whereas in the early period the endings clearly distinguished the various cases, after the
stress accent was introduced the endings were weakened and some were lost. As one result, the root
noun inflection was virtually eliminated.
But as endings were lost, markers for case and number evolved from what were originally
derivational suffixes. New inflections developed from these, such as the ye/yo-stems and especially
the n-stems. Loss also contributed to a reduction of case forms. Evidence for the vocative survived
only in Gothic and for the instrumental in a few paradigms, so these forms must have been reduced

in use by late Proto-Germanic. While some scholars have assumed that the locative was the basis of
many Proto-Germanic dative forms, it is more likely, in accordance with Meillet (1964:294), that a
zero-grade form of the dative was the basis of these, so that the locative need not be recognized for
Germanic generally. Similarly, the ablative need not be recognized on the basis of the few adverbial
forms that have been accounted for as reflexes of it; the adverbial endings may have been
independent of the ablative case, as supported by semantic evidence.
The changes in inflection of nouns were paralleled by those in pronouns, although these were more
conservative in the retention of inherited categories. Both of these influenced the inflection of
adjectives, as will be noted below.
3.4. Inflection of Pronouns
From comparison of the forms in the various Indo-European dialects, it is obvious that in early
Proto-Indo-European there was no pronominal inflection paradigm, but instead a series of forms.
For example, the base changes from the nominative to that of oblique forms, e.g. I vs. me, and what
is more, the form of some of the most frequent pronouns, e.g. that for I, varies from dialect to
dialect, cf. Skt ahm, Gk g. Further, as a complete paradigm of demonstratives was introduced,
the base selected varied from dialect to dialect; for example, the bases in Germanic differ from
those in Italic. As a result of the relatively recent emergence of pronominal inflection in IndoEuropean, etyma are reconstructed here in Proto-Germanic for the pronouns actually attested, rather
than on the basis of reconstructions in Proto-Indo-European.
It is instructive to recall the structural reasons for the infrequency of pronouns in early Proto-IndoEuropean; most important of these is the indication of subjects by verbal endings in the early
language. When verb forms no longer distinguished subjects unambiguously, as when the first and
third singular preterite indicative merged, pronouns came to be used more frequently. Similarly, as
the affixes marking nominal cases merged, agreement was indicated by pronouns, and subsequently
also by pronominal adjectives, which developed full inflections in the individual Indo-European
dialects.
Moreover, the development of a strongly structured paradigm of the noun led to a similar paradigm
for pronouns, in which various particles were added, such as ge in Greek (e)mge and ProtoGermanic mek(e) in the accusative, and by the "freezing" of a form in the genitive of personal
pronouns, and so on.
The non-paradigmatic set of pronouns in Indo-European can be understood most readily from
scrutiny of the Hittite pronoun, which was largely clitic. Although pronouns in Germanic maintain
traces of their former clitic uses, in that they generally occupy positions in the sentence that do not
carry primary stress, they may also be fully stressed. By the time copious material in the dialects
becomes available, pronouns are among the most frequent forms, with fully developed paradigms;
the forms reconstructed here for Proto-Germanic are based on these, even though they may be
relatively recent, having become regularized as the pronominal paradigms emerged.
3.4.1. Personal Pronouns

Personal pronouns are those without a noun as antecedent, as a consequence of which they lack
gender. A full inflection, including dual forms, existed for first and second person; in addition the

third person had three singular forms, used as reflexives, also for first and second person.
First Person Pronouns

1st Person
PGmc
Go.
ON
OE
OHG
Nom.sg.
eka
ik
ek
ic
ih
Gen.sg.
mn
meina
mn
mn
mn
Dat.sg.
miza
mis
mr
m
mir
Acc.sg.
mike
mik
mik
mec, m
mih
Nom.pl.
weys
weis
vr
w
wir
Gen.pl.
unser
unsera
vr
ser, re
unsr
Dat.pl.
uns
uns(is)
oss
s
uns
Acc.pl.
uns
uns(is)
oss
s(ic)
unsih
Nom.du.
wet
wit
vit
wit
Gen.du.
unker
ugkara
okkar
uncer
unker
Dat.du.
unke
ugkis
okkr
unc
Acc.du.
unke
ugk(is)
okkr
unc(it)
The nominative singular posited is probably a reflex of PIE egom. In Runic inscriptions, besides the
simple form ek, variants are attested, as in hate-ka 'my name is' (Lindholm) and haite-ga
(Kragehul). The Kragehul form may be related to the Indo-European form with -gh, cf. Skt ahm.
The Germanic variation in vowel reflects forms with primary stress, ON ek, and those with lesser
degrees of it, OE ic.
The genitive singular form is based on the possessive adjective extended by a suffix, the source of
which is unclear. In the dative singular PIE me was extended by a suffix that cannot be determined
with assurance, possibly -so as in Umbrian seso. As noted above the accusative forms are reflexes
of PIE me to which a particle -ge was added. Old English m is taken over from the dative.
The nominative plural is based on PIE wey-, cf. Skt vay-m 'we', to which the plural suffix was
added in Germanic. In the genitive plural of the dialects other than Old Norse, the adjectival suffix
-ero- was added to the base ns-, and the final -o was lengthened in parallel to that of the genitive
singular. In Old Norse the suffix was added to the nominative base we(y)-. The different forms
indicate the lateness of the formation. In the dative and accusative plurals, -is was extended from
the dative in Gothic to the reflex of PIE ns-, cf. Skt nas, and -ih from the accusative in Old English
and Old High German. These forms are instructive in indicating how Germanic pronominal forms
were expanded.
The nominative dual form is based on PIE wed; similar reflexes are found in OCS v and Lith. vdu
'both of us'. Forms of the dual are not attested in Old High German, apart from one isolated
instance, but reflexes of them survive in the contemporary Bavarian second plural: nominative os,
genitive enker, accusative enk. The genitive dual has the same affix as the genitive plural. The
dative and accusative dual endings were extended in Gothic and Old Norse by that of the dative
plural.
Second Person Pronouns

The second singular forms are remarkably parallel to those of the first singular in development.
Only the forms with special change will be discussed.

2nd Person
PGmc
Go.
ON
OE
OHG
Nom.sg.

d, du
Gen.sg.
n
eina
n
n
dn
Dat.sg.
eza
us
r

dir
Acc.sg.
eke
uk
ik
e(c)
dih
Nom.pl.
ys
jus
r
g, ge
ir
Gen.pl.
izwer
izwara
y(u)ar
ower
iuwr
Dat.pl.
(w)izwiz
izwis
yr
ow
iu
Acc.pl.
(w)izwiz
izwis
yr
ow(ic)
iuwih
Nom.du.
yut
jut
it
git
Gen.du.
inker
igqara
ykkar
incer
Dat.du.
inke
igqis
ykkr
inc
Acc.du.
inke
igqis
ykkr
inc(it)
In the accusative singular the -ge particle was added to te, as in the first person; in Gothic the vowel
of the nominative was extended to the accusative.
For the nominative plural compare Avestan ys, Lith. js. The vowel was modified after that of the
first person in the dialects other than Gothic. In the dative and accusative plural forms PGmc wiz <
PIE wes was reduplicated, possibly to distinguish it from the first person plural nominative. The
initial w was lost by dissimilation, with *iRwiR > *iwiR > yr in Old Norse. In pre-Old English
and Old High German *izwis > *iwis and thereupon the forms attested.
In the dual, the vowel of the nominative was modified after that of the first person pronoun in the
dialects other than Gothic, while the oblique cases have the same suffixes as the first person
pronouns.
The Reflexives

As in the other Indo-European dialects, the reflexive was used for all numbers and genders in the
third person. It was not maintained, however, in Old English. For the first and second person, forms
of the personal pronoun were employed, e.g. mik. For the third person PIE se- was extended with
the endings of the first and second singular pronouns. The accusative was taken over for the dative
in Old High German.
Gen.
Dat.
Acc.

PGmc
sn
sezo
seke

Go.
seina
sis
sik

ON
sn
sr
sik

OHG
sn
sih
sih

3.4.2. Demonstrative Pronouns

The most widespread demonstratives as well as definite articles in the Indo-European dialects are
based on *so-/to-. Forms from *so- occur only in the masculine and feminine nominative singular;
forms from *to-, elsewhere. These roots were originally sentence connectives, as application of
their reflexes in Hittite and to some extent in Sanskrit show. In late Proto-Indo-European and early
Proto-Germanic, endings were added to them and their forms were modified by analogy, especially
in the Germanic dialects.

The demonstrative constructed on Proto-Germanic *sa-, *a-

Masc.Nom.sg.
Masc.Gen.sg.
Masc.Dat.sg.
Masc.Acc.sg.
Masc.Nom.pl.
Masc.Gen.pl.
Masc.Dat.pl.
Masc.Acc.pl.
Neut.Nom/Acc.sg
.
Inst. (or Dat.) Sg.

PIE
so
teso
tosmai

Go.
sa
is
amma

tam
toy
toyzm
toymiz
tonz
tod

PGmc
sa
eza
am(m)/

an(n)
ay
ayzn
aymiz
anz
at(m)

tei

(i)/

ON
s
ess
eim

OE
s
s
m

OHG
der
des
demu

ana
i
iz
im
ans
ata

ann
eir
eira
eim

at

one

ra
m

den
d, die
dero
dm
d, die
daz

Dat. v,

au

Dat. y

diu

Neut.Nom/Acc.pl
t

diu
.
Fem.Nom.sg.
s
s
s
s
so
diu
Fem.Gen.sg.
tess
ezz
izs
eirar
re
dera
Fem.Dat.sg.
tesy
ezay
izi
eire
re
deru
Fem.Acc.sg.
(n)

dea, dia
Fem.Nom.pl.
ts
s
s
r

deo
Fem.Gen.pl.
tesm
ezn
iz
eira
ra
dero
Fem.Dat.pl.
toymiz
aymiz
im
aim
m
dm
Fem.Acc.pl.
ts
s
s
r

deo
The Proto-Germanic forms are fairly transparent; many are direct reflexes of the Proto-IndoEuropean forms. Some have modifications, like the masculine accusative singular, but the principal
modifications have been in the dialects. Regularization is most obvious in the Old High German
forms. These details must be left to the treatments of the individual dialects.
3.4.3. Demonstratives with Further Extensions; Relative Pronouns

In keeping with the archaic position of Proto-Germanic, particles were appended to demonstratives
to create relative pronouns, although different dialects used different particles in this role. In Gothic
a particle -h, -uh was added throughout the paradigm, yielding sah, sh, atuh, etc. Although the
source of the -u is unclear, -h is derived from PIE ke, which is reflected in Latin as -que, where it is
found in compound pronouns, e.g. quisque 'whoever'. The Old Norse paradigm sia, si, etta
comprises several such extensions; the form si illustrates one extension, -a, subsequently
lengthened, presumably the same particle as that in Gothic ana < PGmc ann.
In Old Norse and the West Germanic dialects, the particle -se yielded the most widespread
compound, with the two components clearly evident in Runic sasi, susi, atsi where only the first
component is inflected. The second (-se) component is also found in Lat. ipse 'this, he'. As in Latin,
the Germanic -se component eventually came to be inflected. The early form *e-se is reflected in
OE nom.sg.masc. es, but the oblique case forms isses, issum, isne show final inflections.
Detailed analysis of the development of this compound to the widely used this of English, dieser of
German, and so on, is a concern of the individual dialects.

The relative pronoun in Gothic also shows compounding with the particle -ei added to the
demonstrative, as in saei 'which, that'. The particles es and at are used to indicate the relative
pronoun in Old Norse; in the West Germanic dialects, on the other hand, reflexes of PIE te/o- are
the basis of the relative pronoun, e.g. OE e, OS the, OHG the, de, thie. These illustrate further that
various particles were used in Proto-Germanic to indicate demonstrative and relative reference, so
that no specific pronouns can be reconstructed for the early language for these categories.
3.4.4. The Anaphoric Pronoun

A demonstrative based on various pronominal stems that is less emphatic than forms based on
Proto-Germanic so is found in the dialects. In Gothic and Old High German this is es, er in the
nominative singular masculine, in contrast with OE he, OS he, ON hann. Due to such diversity, a
complete paradigm cannot be reconstructed for Proto-Germanic; instead, forms from various roots
were selected, and complete paradigms were constructed in the dialects around one or more of
these, with endings in general taken from the *so paradigm. The roots are:
PIE ei-, i-; cf. Lat. is 'he', Go. is
PIE sye-, sy-; cf. Skt sya, sy 'that', OHG sie
PIE to-; cf. Gk t, ON at, eir
PIE ke/o- (cf. Oscan e-kas 'these') was extended by -eno-, producing*hno, from which ON hana
and its paradigm developed. As the variety illustrates, a paradigm cannot be reconstructed for ProtoGermanic, but instead a number of pronominal forms were in anaphoric use.
3.4.5. The Interrogative Pronoun

Forms based on the stem PIE ko-, key- are found throughout the Indo-European dialects for the
singular, although there are no distinct forms for the feminine. The forms in the Germanic dialects
have undergone considerable modification, by analogy with forms of demonstratives. Moreover,
Gothic includes the feminine forms nominative & accusative ho, dative hizai, which are probably
patterned on the demonstrative sa rather than on reflexes of Proto-Indo-European k. These attest
further to the regularizing of paradigms in the individual dialects. The masculine (animate) and
neuter forms are as follows:
PIE
kos
keso
kosmai
kom
k/key
kod

PGmc
was
weza
wamm
wana
w/w
hwat

Go.
has
his
hamma
hana
h
ha

ON
(huerr)
hues
hweim

OE
hw
hws
hwm
hwone
hw, hwy
hwt

OHG
hwer
hwes
hwemu
hwenan
hwiu
hwaz

Masc.Nom.
Masc.Gen.
Masc.Dat.
Masc.Acc.
Masc.Inst.
Neut.Nom/Acc.
huat
Neut.Dat.
hw
In the nominative, the Gothic form is a regular reflex of the Proto-Indo-European form, as are the
neuter forms. In the Old English form the final consonant is lost, with subsequent lengthening of the
final vowel; in the Old High German form the vowel of the oblique cases has been introduced. In
Old Norse a compound form has been introduced. The forms of the other cases are patterned after
those of the demonstrative except in the instrumental and the Old Norse dative.

As in the demonstrative, various compound forms of interrogatives are attested. In Gothic and Old
Norse a compound interrogative is based on the interrogative particle Go. har, ON huar 'where'
plus a reflex of PGmc -yos, e.g. Go. harjis, ON huerr 'who'; it is inflected like a strong adjective.
The form huerr has replaced the simple interrogative in the Old Norse nominative.
Reflexes of PGmc hwe/a-ter- are found in all the dialects. The West Germanic forms are based on
the -e- form, OE hweer, OHG hwedar 'which of two'; the forms in the other dialects are based on
the -a- form, Go. haar, ON huaarr. A reflex of the ko- form is also attested in OE hwer,
suggesting that in the other dialects one form was generalized throughout the paradigm.
Reflexes of a compound with PGmc lko- 'body' as second component are attested through all the
dialects, but with varying first elements: PIE ki in Go. hileiks, OE hwilc; PIE key- in ON hvlkr;
PIE ko- in OE hwelc, OHG hwelih. The vowel in question resulted from regularization in the
dialects.
In addition, various indefinites derived from interrogatives, e.g. Go. hazuh 'each', OE ghw
'each', OHG etewer 'someone', vary from dialect to dialect, hence a common Proto-Germanic
etymon cannot be reconstructed.
3.5. Inflection of Adjectives
One of the characteristics of Germanic is the development of two adjectival inflections: one
preserves the inflections assumed for late Proto-Indo-European; the other is a new inflection based
on n-stems. As one of their uses in the proto-language, n-stems referred to specific individuals, as in
Latin personal names like Cat, Catnis literally 'the wise one'. The n-inflection of adjectives,
called "weak" by Jacob Grimm in contrast with the inherited inflection called "strong," indicated a
specific item in accordance with the Proto-Indo-European meaning, much as the definite articles did
later.
3.5.1. The Strong Inflection of Adjectives

It is assumed that adjectives were originally inflected like nouns in Proto-Indo-European; the reflex
of this inflection in Germanic is the strong declension. Both have the same stems, consonantal and
-i-, -o-, -u- as well as -yo-. But as adjectives were frequently used in conjunction with demonstrative
pronouns, some inflections were taken over from these. It is unclear how many of the endings were
so modified in early Germanic. When all the dialects agree on a pronominal ending, as in the
genitive, dative and accusative singular masculine, as well as in the plural forms and the
nominative/accusative singular neuter, they may have been inherited from the parent language. In
the feminine Gothic is more conservative, maintaining noun endings except in the genitive singular,
and the genitive and dative plural.
The following inflection of *halbs 'half' may be assumed for late Proto-Germanic, with reflexes in
the dialects.
Masc.Nom.sg.
Masc.Gen.sg.
Masc.Dat.sg.
Masc.Acc.sg.

PGmc
halbaz
halbas
halbazmai
halbanm

Go.
halbs
halbis
halbamma
halbana

ON
halfr
halfs
hlbom
halfan

OE
healf
healfes
healfum
healbne

OHG
halb
halbes
halbemu
halban

Masc.Nom.pl.
Masc.Gen.pl.
Masc.Dat.pl.
Masc.Acc.pl.
Neut.Nom.sg.
Neut.Nom/Acc.pl.
Fem.Nom.sg.
Fem.Gen.sg.
Fem.Dat.sg.
Fem.Acc.sg.
Fem.Nom.pl.
Fem.Gen.pl.
Fem.Dat.pl.
Fem.Acc.pl.

halbay
halbaysam
halbaymis
halbanz
halbatam
halb
halb
halbaizs
halby
halbon
halbz
halbaiz
halbaimiz
halbz

halbi
halbiz
halbim
halbans
halb(ata)
halba
halba
halbizs
halbi
halba
halbs
halbiz
halbaim
halbs

halber
halbra
hlbom
halba
halbt
hlb
hlb
halbrar
halbre
halba
halbar
halbra
hlbom
halbar

healbe
healbra
halbum
halbe
healf
halfu
halbu
halbre
halbre
halbe
halba, -e
halbra
halbum
halba, -e

halbe
halbero
halbm
halbe
halb
halbu
halb
halberu
halberu
halba
halbo
halbero
halbm
halbo

3.5.2. The Weak Inflection of Adjectives

The basis of the endings was presented in section 3.3.2 above.

The n-stems

Masc.Nom.sg.
Masc.Gen.sg.
Masc.Dat.sg.
Masc.Acc.sg.
Masc.Nom.pl.
Masc.Gen.pl.
Masc.Dat.pl.
Masc.Acc.pl.
Neut.Nom/Acc.sg.
Neut.Nom/Acc.pl.
Fem.Nom.sg.
Fem.Gen.sg.
Fem.Dat.sg.
Fem.Acc.sg.
Fem.Nom.pl.
Fem.Gen.pl.
Fem.Dat.pl.
Fem.Acc.pl.

PGmc
halb
halbenaz
halbeni
halbonun
halbaniz
halbanm
halbammiz
halbunz
halb
halbn
halbn
halbn(i)z
halbni
halbnum
halbniz
halbnam
halbamm(iz)
halbnz

Go.
halba
halbins
halbin
halban
halbans
halban
halbam
halbans
halb
halbna
halb
halbns
halbn
halbn
halbns
halbn
halbm
halbns

ON
halbe
halba
halba
halba
hlbu
hlba
hlbom
hlbu
halba
hlbo
halba
hlbo
hlbo
hlbo
hlbu
hlbu
hlbom
hlbo

OE
halba
halban
halban
halban
halban
halbena
halbum
halban
halbe
halban
halbe
halban
halban
halban
halban
halbena
halbum
halban

OHG
halbo
halben
halben
halbon
halbon
halbno
halbm
halbon
halba
halbun
halba
halbn
halbn
halbn
halban
halbno
halbm
halbn

3.5.3. The Comparison of Adjectives

Adjectives are also inflected in comparative and superlative degrees. The formations are treated in
derivational morphology, section 4.8
3.6. The Numerals
The first three cardinal numerals are inflected; the cardinals are listed in section 6.6.1. The ordinal
numerals are inflected in the weak declension.

3.7. Inflection of Verbs

3.7.1. Origin of the Tense System

As noted above, verbs are inflected for tense, mood, number, and person. They also have nominal
forms, the infinitive and participles. Four forms of the bases are listed as the principal parts, as
illustrated here by the Old High German forms of the verb 'to see': infinitive sehan, preterite 1/3
singular sah, preterite 1 plural shum, past participle gisehan. From these, presumably all forms of
the verb can be identified or produced.
There are two major classes of verbs, as determined by their inflection for tense. The class in which
the inflections for tense are indicated by a change of vowel in the base is called "strong," after a
term introduced by Jacob Grimm; sehan, sah is an example. The other class, where differences of
tense are indicated by a dental suffix, is called "weak," also after Grimm; because the same base is
used throughout the preterite, only three principal parts are given, as for OE legan, legde, gilegd
'lay'. Both classes are inflected in two moods, indicative and subjunctive, which are distinguished
by endings, as are also singular and plural number, first, second and third person. Verbs also have an
imperative, and medio-passive forms have survived in the present of Gothic. Paradigms for these
are given below.
In previous grammars the inflections have been related to those of Sanskrit, Greek, and Latin, as by
Prokosch (1939:147-159). Since the discovery of Hittite, on the other hand, it has become clear that
the Germanic inflections are continuations of the pattern in Proto-Indo-European; the more complex
inflections in Sanskrit, Greek and Latin, on the other hand, are later developments in these dialects.
The Germanic verb system was formed when the early language still maintained characteristics of
Active language structure. Active languages distinguish three classes of verbs: active/animate,
stative/inanimate, and a small class of involuntary or impersonal verbs. Characteristic verbs of these
classes were pointed out in 1897 by Delbrck, but he did not associate them with an earlier
structural type (Gdr. 4:178-213, 417-478). Among active verbs he listed those for 'eat', 'bite', 'creep';
among stative verbs he listed those indicating joy, sorrow, satisfaction (cf. Lehmann 1993:218-223,
2002:77-81). The small class has three subsets: verbs referring to natural phenomena like raining;
verbs referring to psychological states or conditions, like being disgusted; verbs referring to
necessity, obligation or capability, like 'it is necessary'. Moreover, lacking a verb for 'have', the
meaning was expressed in the proto-language with the pronoun in the dative and the third singular
of the verb 'to be', as in Latin mihi est 'to me is = I have'.
When the tense system was introduced, the two large classes were combined to make up the
Germanic conjugation, in which the preterite originally indicated a state, as do the Hittite preterite
and originally the Greek and Sanskrit perfect. The impersonal verbs were the basis of the preteritepresents, so named because they arose from preterite forms but came to have present meaning.
3.7.2. The Strong Verb System

The arrangements of the strong verbs in seven classes reflect the earlier distinction. In verbs of the
first five classes, the present tense forms are continuations of active inflection in accordance with
their active/animate meanings, e.g. Class I Go. steigan 'climb', Class II kiusan 'choose', Class III
hilpan 'help', Class IV niman 'take, accept', Class V lisan 'read'; the preterite is based at least in part
on the perfect of Proto-Indo-European, which indicated state as the result of completed action.

Unlike the first five strong verb classes, the sixth and seventh have the base form in the preterite.
The evidence for this arrangement in the sixth and seventh classes was recognized by Prokosch
(1939:150-151), following Brugmann and Wood, but was not explained. The basis is now clear.
Many verbs of these two classes indicate a state rather than action, e.g. standan 'stand', haitan 'call,
be called'. Their meanings then corresponded to that of the early Indo-European stative, and
accordingly their base form was used to make the Germanic preterite rather than the present, e.g.
PIE st-, Go. st 'stood', PIE kyd-, OE ht 'was called'; they then formed a new present tense
form, as illustrated with the infinitives above.
This explanation receives support from the weak preterite, which was based on the addition of a
*dh-suffix that indicated state (Lehmann 1943). It receives additional support from verbs that
maintained the inflection of the perfect (stative) but shifted their meaning, such as Gk oda, Go.
wait 'I know' from the base *weid- 'see'; rather than retention as preterite of the root, the form
shifted in meaning from 'I have seen' to 'I know'. Further discussion is included below, as in
treatment of the sixth and seventh classes of the strong verbs.
3.7.2.1 The First Five Classes of Strong Verbs

The five classes are traditionally labeled by the vocalism of their bases. The first class has a base
vowel -ei- plus a consonant, as in the principal parts of the Gothic verb steigan, staig, stigum,
stigans 'climb'. The second class has a base vowel -eu- (Gothic i < e) plus a consonant, as in the
principal parts of the Gothic verb kiusan, kaus, kusum, kusans 'choose'. The third class has a base
vowel -e- and one of the resonants l, m, n, r plus a consonant, as in the principal parts of the Gothic
verb hilpan, halp, hulpum, hulpans 'help'. The fourth class has a base vowel -e- followed by one of
the resonants, as in the principal parts of the Gothic verb stilan, stal, stlum, stulans 'steal'. The fifth
class has a base vowel -e- followed by one of the other consonants, as in the principal parts of the
Gothic verb lisan, las, lsum, lisans 'read'.
3.7.2.2 The Sixth and Seventh Classes of Strong Verbs

The original verbs of the sixth and seventh classes have, as noted above, the base form in the
preterite because their meaning was stative. Since new forms were created for the present, and these
vary, they must be noted individually. The forms of the seventh class verbs have many special
problems. The base form in verbs of the sixth class is parallel to that of the fourth and fifth classes,
but the vowel was long, consisting originally of -e- plus a laryngeal, as in the base given as PIE st< stex- 'stand', cf. Gk hstmi, hstmi. In Germanic this was followed by *nt > nd and , as in the
principal parts of the Gothic verb standan, st, stum standans. The base form in most of the
verbs of the seventh class is parallel to that of the first three classes, again with long vowel, as in the
principal parts of the Old High German verbs heizan, hiaz, hiazum, giheizan 'call', loufan, lief,
liefum, giloufan 'leap', and rtan, riet, rieten, giratan 'advise'.
3.7.3. The Four Classes of Weak Verbs

The four classes of weak verbs are distinguished by their suffixes, and also by their meaning. Class
1 has a -ja- suffix based on PIE -ye/o- with -a- from PIE -o- in the root, as in PGmc lagjan 'lay' in
contrast with PGmc ligan 'lie'; as in this verb they have causative or factitive meaning. While many
in this class are based on verbal roots, others are based on nominals, such as hailjan 'heal'. Class 2
has a suffix based on -- from PIE -- as in salbn 'anoint' and are chiefly denominatives; cf. OHG
salfs 'a salve'. Class 3 has an -i- suffix, as illustrated by the Go. preterite habida (cf. the infinitive

haban) and OHG habta, though not in the other dialects, as in OE hfde 'had'. Class 4 has a -nsuffix based on PIE -n-, as in Go. waknda, with a shortened form in ON vaknae 'wakened'; the
vowel was also weakened in the present, as in Go. wakna, ON vakna 'I waken'. Verbs of the class
have inchoative or middle meaning.
3.8. The Inflected Forms
The present tense inflected forms may be illustrated by the Proto-Germanic verb neman, forms of
which are among the best attested in Gothic, including the dual, and also in the other dialects.
Gothic forms are listed for comparison:
Proto-Germanic
Indicative Subjunctive Imperative
1 sg. nem
nemai
2 sg. nemes
nemais
nem
3 sg. neme
nemai
nemadu
1 du. nems
nemaiwa
2 du. nemats
nemaits
nemats
1 pl. nemam
nemaima
2 pl. neme
namai
nemi
3 pl. nemand
nemaina
nemandu
The forms of the preterite are as follows:

Gothic
Indicative Subjunctive
nima
nimu
nimis
nimis
nimi
nimi
nims
nimiwa
nimats
nimits
nimam
nimima
nimi
nimi
nimand
nimina

Imperative
nim
nimadu
nimats
nimi
nimandu

Proto-Germanic
Gothic
Indicative Subjunctive
Indicative Subjunctive
1 sg. nam
nmjai
nam
nmju
2 sg. namt
nms
namt
nmeis
3 sg. nam
nmi
nam
nmi
1 du. nmu
nmu
2 du. nmuts
nmts
nmuts
nmeits
1 pl. nmum
nmma
nmum
nmeima
2 pl. nmu
nm
nmu
nmei
3 pl. nmun
nmna
nmun
nmeina
The forms of the weak verbs are comparable to those of the strong verbs in the present, although the
suffixes must be noted. Proto-Germanic examples in the present singular are as follows (with the
plural forms being like those of the strong verbs):
Class 1
Class 2 Class 3 Class 4
1 sg. lagja
skja salb
haba
wakna
2 sg. lagjes sks salbs
habis
wakns
3 sg. lagje sk salb habi wakn
Singular forms of the weak verb preterite are as follows (with the plural forms being like those of
strong verbs):
Class 1 Class 2
1 sg. lagida
salbda
2 sg. lagids salbds

Class 3
habida
habids

Class 4
waknada
waknads

3 sg. lagida
salbda habida waknada
Forms of the passive present have also been attested in Gothic, where the subjunctive endings are
based on the innovated -u in the first singular. Only occasional examples are attested in the other
dialects.
Proto-Germanic
Gothic
Indicative Subjunctive
Indicative Subjunctive
1 sg. lagjada
lagjaidai
lagjada
lagjidu
2 sg. lagjaza
lagjaizai
lagjaza
lagjizu
3 sg. lagjada
lagjaidai
lagjada
lagjidu
1-3 pl. lagjanda
lagjaindai
lagjanda
lagjindu
There are three non-finite forms: the infinitive, e.g. neman, lagjan; the present participle, e.g.
nemands, lagjands; the past participle, e.g. numans, lagis. The present participle is inflected like a
weak adjective; the feminine nominative ends in -, e.g. nemand, lagjand.
3.9. The Preterite-Presents
As noted above, the third class of verbs as developed from the stative stage has preterite forms with
present meanings, and in accordance with the emphasis on morphology in the 19th century it was
labeled "preterite-present." As with wit, earlier 'I have seen', their basic meaning in the IndoEuropean perfect was stative but corresponded to an activity in the present rather than in the past.
Similarly, PGmc dars corresponds to Gk thars 'I am courageous'; from a meaning corresponding
to 'I am in a state of being courageous', the meaning 'I dare' developed.
The preterite-present verbs are inflected like the first six classes of strong verbs. Because the
endings are like those of the strong verb preterite, only the first singular and first plural present, and
also the first singular preterite and the past participle are given here, listed by classes.
Some of them are found in other Indo-European dialects as well, notably the equivalent of wit: Skt
vda, Gk oda 'I know'. These are o-grade forms of the Proto-Indo-European base weid- 'see'. It is
instructive to note that the Lat. vd 'I have seen' was continued as the regular perfect form of vide
'I see'. In Sanskrit, Greek and Germanic, on the other hand, the completed action was treated as a
state, and then shifted semantically to 'I know'. The shift in the other verbs is comparable, as is clear
from their vocalism. The principal parts of Gothic preterite-present verbs are as follows:
1
2
3

4
5
6

Pres. Sg.
wait
ih
dug
kann
ann
arf
ga-dars
skal
man
ganah
mag

Pres. Pl.
witum
ihum
dugum
kunnum
unnum
urfum
ga-darsum
skulum
mumum

Pret.
wissa
ihta
duhta
kuna
una
arfta
ga-dursta
skulda
munda

magum

mahta

Past Ptc.
Gloss
wiss
'know'
ihan
'have'
'it suffices'
kuns
'know'
unnan
'love'
arft
'have need of'
'dare'
skuld
'have to'
mund
'believe'
binahts 'is adequate'
maht
'be able'

g
ga-mt

gum
mtum

hta
ga-mhta

'fear'
'find'

3.10. The Uses of the Forms

The uses of the indicative tense verb forms are comparable to those in the current Germanic
languages, which mark the basic contrast between present time and past time. Present tense forms
are also employed to express future time; but compound forms with auxiliaries gradually replace
them, for the most part, in the dialects.
There may be some retention of aspect, especially in the preterite forms.
The subjunctive forms are used to indicate uncertainty, as well as in some subordinate clauses.
Fuller treatment will be given in the section on the structure of sentences.
The uses of the imperative, the infinitive and the participles are comparable to those in the current
languages.
The uses of the nominal forms are comparable to those in the current Germanic languages. As
exemplified above, duals are maintained in some inflections; they invariably represent two entities,
so that the plurals are less coverable in the relevant paradigms.