4. Nouns

4.1. Declension of Nouns

4.1.1. Declension is made by adding terminations to different stem endings, vowel or consonant.

Adjectives are generally declined like nouns, and are etymologically to be classed with them, but they have some peculiarities of inflection which will be later explained.

4.1.2. There are eight commonly reconstructed cases for Late Indo-European:

I. The nominative is the case of the subject of a sentence and predicate nominative.

II. The vocative is the case of direct address.

III. The accusative is the case of the direct object of a verb.  It is used also with many prepositions.

IV. The genitive may generally be translated by the English possessive.

V. The ablative, the source or place from which.

VI. The dative, the case of the indirect object. It also indicates possession, and beneficiary of an action.

VII. The locative, the place where.

VIII. The instrumental, the means and the agent.

NOTE. The oblique cases appear in the English pronoun set; these pronouns are often called objective pronouns; as in she loves me (accusative), give it to me (dative) or that dirt wasn’t wiped with me (instrumental), where me is not inflected differently in any of these uses; it is used for all grammatical relationships except the genitive case of possession, mine, and a non-disjunctive nominative case as the subject, I.

4.1.3. Nouns and adjectives are inflected in LIE in four regular declensions, distinguished by their final phonemes – characteristic of the stem –, and by the opposition of different forms in irregular nouns. They are numbered following Graeco-Roman tradition: first or a-stem declension, second or o-stem declension, third or i/u-stem declension, fourth or c-stem declension, and the variable nouns.

The stem of a noun may be found, if a consonant stem, by omitting the case-ending; if a vowel stem, by substituting for the case-ending the characteristic vowel.

Decl.

Stem ending

Nominative

Genitive

1.

ā (ja/ī, ē, ō)

-

-s

2.

e/o (Thematic)

m., f.-s, n.-m

-os/-osjo

3.

i, u and Diphthong

m., f.-s, n.-

-eis, -eus; -jos, -wos

4.

Resonants & Consonants

-s, -

-es/-os

(5)

Heteroclites

-, -r

-(e)n

NOTE. Most Indo-Europeanists tend to distinguish at least two major types of declension for the oldest PIE, thematic and athematic. Thematic nominal stems are formed with a suffix -o- (also -e-), and the stem does not undergo ablaut, i.e. there is no ablaut difference between the strong and the weak cases, and there is no accent change, see below.

Feminine stems in -ā < *-eh2 were originally c-stems with final *-h2 which, under pressure from the o-stem adjectives, were adapted to the thematic paradigm of the masculine o-stems. It is sometimes separated from the athematic declension into a new class, even if being originally consonantal, because of such peculiarities; so e.g. in Clackson (2007): “In respect of the reconstructed case-endings, the class of feminine nouns in [a-stem declension] shows clear affinities with the athematic class [c-stem declension], and the o-stem declension diverges more radically from both. In the daughter languages, however, there is a general tendency for the o-stem class and the feminine ā-stems to become more closely associated, almost certainly through the combination of the two classes in a number of pronominal and adjectival declensions as masculine and feminine alternatives.”

The declension of i/u-stems and c-stems is more complex, as it involves accent-ablaut changes. Fortunately the most productive (thus more frequent) declensions in LIE are mainly o-stems and ā-stems.

NOTE. Stems in consonant, i, and u, are more archaic, and they are classified further by their ablaut behaviour into different so-called dynamic patterns, after the positioning of the early PIE accent in the paradigm. See below §4.7.

The distinction of i-stems and u-stems from c-stems is also traditional, but according to Fortson (2004), to keep that distinction for PIH probably “is both unnecessary and misleading, as it masks the fundamentally identical behavior of both groups over against that of the thematic nouns”. In LIE, however, there are pragmatic reasons to distinguish them.

4.1.3. The following are general rules of declension:

a. The nominative singular for animates ends in -s when the stem endings are i, u, diphthong, occlusive and thematic (-os); in - when the stem ends in ā, resonant and s; in the plural -es is general, -s for those in ā, and -os for the thematic ones.

NOTE. For collectives/feminines in -ja/-ī <*-jə<*-ih2 we prefer to use in our texts the ending -ja for feminines, and -ī for neuters as a general rule. It is not intended as a ‘normative’ selection, though, but as a conventional simplification of the otherwise tedious repetition -ja/-ī that is followed in other books, while at the same time reflecting the natural evolution pattern of such forms in NWIE (see below §§4.2, 4.4.2). So e.g. this convention does not limit the use of feminines in -ī; as e.g. Lat. -trīx, or coxendīx.

b. The accusative singular of all masculines and feminines ends in -m or - (after consonant), the accusative plural in -ns or -ṇs.

NOTE. A general accusative plural ending -ns (-safter consonant) is usually reconstructed for Late Indo-European, because e.g. within the u-stem from PIE *-u-ns, early IE languages show *-uns, *-ūns, *-ūs; cf. Goth. sununs, O.Ind. sūnū́n, Gk. uíuns, Lith. sū́nus, O.C.S. syny, Lat. manūs. See H. Rix (FS Risch 1986). Most scholars also posit an ‘original’, older **-ms form (a logical accusative singular -m- plus the plural mark -s), but they usually prefer to reconstruct the attested *-ns, thus implicitly suggesting either the theoretical origin of the ending, or a previous PIH *-ms LIE *-ns. For a PIH *-ms, cf. maybe Hitt. -uš (Ottinger 1979), but Lyc -s (<*-ns?). To be consistent with decisions taken elsewhere in this grammar (as e.g. reconstructed PIE *-T(s)T- as NWIE intermediate -sT-, see §2.7), the intermediate, attested -ns is the conservative choice, whereas **-ms is just a probable hypothesis about its actual origin.

c. The vocative singular for animates is always -, and in the plural it is identical to the nominative.

d. The genitive singular is common to animates and inanimates, it is formed with -s: -s, -es, -os. An alternative possibility is extended -os-jo. The genitive plural is formed in -ōm, and in -ām in a-stems.

NOTE. Case endings in -e/o-, -ē/ō-, are generally written in this book in -o-, -ō-, in inflected nouns, given the alternating nature of these forms even within the same dialectal branches, and the unknown nature of the original ablauting forms. Sihler (1995), Fortson (2004), Ringe (2006), reconstruct ‘original’ forms in -es, while Beekes (1995) deems -(o)s the oldest athematic declension. Meier-Brügger (2003) or Adrados–Bernabé–Mendoza reconstruct both as alternating -es/-os. For Sihler (1995), since e-grade is typical of consonant stems, maybe forms in -o- in o-stems were the ‘original’ ones, and those in -e- were secondary creations; that same argument is found for the genitive of athematic nouns in -es, supposedly substituted by the ‘original’ thematic -os (Fortson 2004). In any case, it seems that in Late Indo-European forms in -o- prevailed, as did the thematic declension; as, kunós, over kunés, from kwōn.

e. The obliques singular end usually in -i: it can be -i-, -ei-, -ēi-, -oi-, -ōi- or -āi-, and their extensions. In the plural, there are two series of declensions, instr. -bhis/-mis (from sg. -bhi), dat.-abl. -bhos/-mos (PII -bhjas) as well as (BSl. and PII) loc. in -su (PGk. -si).

NOTE. Comparison shows an ins. sg. -bhi, (cf. Gk. -phi, Myc. -pi, and also Arm. ins. marb), BSl. -mi (cf. Lith. akmenimì, O.C.S. kamenĭmĭ) and for Northwestern IE dialects a division between Italic+Celtic and Germanic+Balto-Slavic Plural forms: Celtic shows traces of an instrumental -bhis (cf. O.Ir. dat.-loc.-ins.-abl. cridib, and in Graeco-Aryan, cf. O.Ind. sūnúbhis, Av. bāzubīs, Arm. srtiwkh), Italic and Celtic show a dat.-abl. -bhos (cf. Celtiberian dat.-loc.-ins.-abl. arecoraticubos, Lat. matribus, Osc. luisarifs), while Balto-Slavic shows Inst. -mis (cf. Lith. sunumìs, O.C.S. synumĭ), dat.-abl. -mos (cf. O.C.S. synŭmŭ, Lith. sūnùms, sūnùmus), and Germanic shows a dat.-abl.-ins. -m-.

Meier-Brügger (2003) considers that “[e]vidence seems to indicate that while the dative and ablative plural were marked with *-mos, the instrumental plural was marked with *-bhi (…) Thus, -bh- would have established itself in Italic and Indo-Iranian as the sole initial consonant, replacing -m-. Conversely, -m- would have established itself in Balto-Slavic and Germanic. Indo-Iranian *-bhi̯as can thus be regarded as a cross between the instrumental *-bhi and the dative/ablative *-mos”. Similarly Mallory–Adams (2007) differentiate for the oldest PIH declension a Dat. *-mus, instrumental *-bhi, and Abl. *-bh(j)os. Kortlandt (1983) and Beekes (1985) reconstruct an original Dat. Pl. *-mus  and Abl. Pl. *-jos, both supposedly substituted later by the ending *-bh(j)os.

For an original *-mus, the Leiden school revitalised an old claim (Van Helten 1890, Loewe 1918) to explain the lack of Umlaut of -i in dat. pl., as e.g. OHG tagum, which followed Georgiev’s (1963) and Kortlandt’s (1983) proposals to explain the ending away by comparing it to O.Lith. -mus, thereby suggesting an original *-u-. That proposal has been questioned e.g. on the basis that the O.Prus. dat. pl. -mans, is generally believed to have resulted from the contamination of dat. Pl. *-mos > -mas, and acc. Pl. -ans (Bemeker 1896, Brugmann 1911, Poljakov 1995). The Leiden position has been questioned on the grounds of the late and dialectal character of BSl. and Gmc. edings among others by Álvarez-Pedrosa (2001, see <http://revistas.ucm.es/fll/15781763/articulos/ESLC0101110239A.PDF>) and Halla-aho (2006, see <http://ethesis.helsinki.fi/julkaisut/hum/slavi/vk/halla-aho/problems.pdf>).

Fortson (2004) sums up the problem of reconstructing -bh- and -m- endings for PIH: “ All this taken together suggests that the *-bh- and * -m- endings developed late, probably after Anatolian split off from the family, and may have originally been postpositions or adverbs ultimately related to Eng. by and Germ. mit ‘with’. (It is cross-linguistically common for postpositions to develop into case-endings.)”

What is certain is that there was a NWIE west/east dialectal differentiation into -bh- or -m-; i.e. dat.-abl. pl. -bhos/-mos (and PII -bhjas), ins. sg. -bhi/-mi, ins. pl. -bhis/-mis. We generally prefer to write the only the forms in -bh- in this grammar, though, given the extension of those forms in all PIE territory, against the forms in -m-, limited to Germanic and Balto-Slavic. In any case, when writing these endings, one should keep in mind that they are dialectally distributed in a uniform way, so that forms in -bh- are not found in the same branches as those in -m-; i.e., if you use forms in -m- when writing or speaking IE, don’t use forms in -bh-, and vice versa.

f. Inanimates have a syncretic form for nom.-acc.-voc. - in athematics, or -m in thematics. The plural forms end in -.

NOTE. Inanimates have a nom.-voc.-acc in *-(e/o)h2, which evolved as -ā in Sanskrit and Slavic, and -a in most other dialects. A convention is therefore followed in this book, using short -a to distinguish the overlapping neu. pl. nom.-voc.-acc. from the  fem. sg. nom.

g. All animates share the same form in the plural for nom.-voc. -es.

4.1.4. The so-called oblique cases - opposed to the straight ones, nom.-acc.-voc –, are the genitive and the obliques proper, i.e. ablative, dative, locative, and instrumental. Straight cases are generally identified with strong cases (those which do not undergo ablaut in athematic declension), while the rest are the weak cases.

NOTE. IE languages show an irregular oblique declension system, especially in the plural, due to its syncretic original nature and to late dialectal merging trends. Sanskrit or Avestan had 8 cases. Anatolian and Italic dialects show up to 8 (cf. Osc. loc. aasai for Lat. ‘in ārā’, or ins. cadeis amnud for Lat. ‘inimicitiae causae’, preiuatud for Lat. ‘prīuātō’, etc.). Balto-Slavic shows seven, Mycenaean at least six cases, while Koiné Greek and Proto-Germanic had five.

Nominal Desinences (Summary)

 

Singular

Plural

 

Animates

Inanimates

Animates

Inanimates

NOM.

-s, -

-m, -

-es

-, -

VOC.

-e, -

ACC.

-m

-ns

GEN.

-es/-os/-s; -osjo

-om

ABL.

-es/-os/-s; -ēd/-ōd/-d

-bhos (-mos); -om

DAT.

-ei

-bhos (-mos)

LOC.

-i

-su (-si)

INS.

-ē/-◌̄; -bhi (-mi)

-eis; -bhis (-mis)

4.2. First Declension

4.2.1. First Declension Paradigm

1. They are usually animate nouns and end in ā (or ), and rarely in ja/ī, ē, and ō.  Stems in ā are very common, generally feminine in nouns and always in adjectives, and the ā ending is used to make feminines in the adjectival motion. Those in ja/ī are rare, generally feminine, and etymologically identical to the neuter plural in nom.-acc.-voc. Those in ō and ē are feminine only in lesser used words.

NOTE. The entire stem could have been reduced to IE a (hence a-stem declension), because this is the origin of the whole PIE stem system, the ending *-(e)h2.

2. The IE first declension corresponds loosely to the Latin first declension (cf. Lat. rosa, rosae, or puella, puellae), and to the Ancient Greek alpha declension (cf. Gk. χώρᾱ, χώρᾱς, or τῑμή, τῑμῆς).

a-Declension Singular Paradigm

NOM.

-

VOC.

- (-◌̆)

ACC.

-m

GEN.

-s

ABL.

-d/(-s)

DAT.

-i

LOC.

-i

INS.

-/-bhi (-mi)

NOTE. This declension in ā, older *-eh2, is usually reconstructed in the singular as older (athematic) PIH nom.-voc. *-eh2 (voc. -h2e?) acc. *-eh2m̥, gen.(-abl.) *-(e)h2os, dat. *-(e)h2ei, loc. *-eh2i, ins. *-(e)h2eh1,  abl. *-(e)h2ed; as, dat. *h1ekw(e)h2ei ekwāi (see Beekes 1995, Clackson 2007). The ablative sg. was linked to the genitive sg. (-s) in the older stages of the language, but as the feminine declension was adapted to the thematic declension in o/e, an ablative in -d was generalised already by Late Indo-European. From Beekes (1995), Adrados–Bernabé–Mendoza (1995-1998), Clackson (2007). 

3. It is therefore identical to those nouns in r, n, s of the fourth declension, but for some details in vocalism: the gen. has an -s and not -es/-os; the difference between nom. and voc. can be obtained by nom. -ā vs. voc. -a (as found in Gk. and BSl.). The zero-grade of the nom.-acc.-voc. in ja/ī stems is different from the gen. in jā.

4.2.2. First Declension in Examples

1. Nominative singular in -; as, ek, mare, patrjā, fatherland, adj. newā, new, cowij, bovine.

NOTE. The representative noun of this paradigm is the word for ‘mare’ which occurs in Sanskrit áś, Latin equa and (Old) Lithuanian aš. This word is probably not of common PIE origin, although it is likely a post-LIE word that appeared to differentiate the feminine of the animate ‘horse’, which was previously used for male and female alike. Clackson (2007).

Examples of ja/ī include potnja/potnī, lady, mistress, deiwja/deiwī, goddess.

NOTE. Stems in ja/ī (<*-ih2) are productive in adjectives of the -u- and  -nt- form, found generally as -ī in Indo-Iranian (cf. Skt. bhárantī), -ja in NWIE (cf. O.C.S. nesǫšti <*-ontj-) and PGk (cf. phérousa<*-ontja), Beekes (1995). Because they were not productive in nouns already in IEDs, the declension of the attested nouns is frozen as an athematic stem from which they derive. So e.g. from nom. deiwī, Skt. dev-, Gk. dī̂-a (<*diw-ja), gen.-abl. diwjs, cf. Skt. dev-ys, Gk. d-ās (<*diw-jās). Clackson (2007).

Those in ē, ō, also rare, make the nominative in -s; as, spekjēs, aspect.

NOTE. These are known from Latin (since Indo-Iranian merged ē with ā especially from the word for ‘path’. Like the forms from *-ih2, these old nouns in *-eH or *-oH retain their c-stem declension paradigm (with accent-ablaut); a famous example found in Beekes (1995) or Fortson (2004) includes pontēs, path, found in Skt. pánthās, Av. pantå, and afield in Lat. vātēs – although later reinterpreted as i-stem pontis in NWIE, cf. Lat. pons, Russ. put’, O.Pruss. pintis. Its declension is reconstructed as nom. pontēs, acc. pontēm, gen. ptós, ins. pl. ptbhí.

2. Vocative singular in -. It is normally identical to the nominative, but disambiguation could happen with distinct vowel grades, i.e. nom. in -ā, voc. in -a.

NOTE. According to Ringe (2006), the vocative of those in ja/ī<*-iH would have been made in *-i, following the example of the ‘fall of the laryngeal’ in vocatives of those in *-aH.

3. Accusative singular in -m; as, ekwām, patrjām, potnjam/potnīm, spekjēm.

4. Genitive singular in -s; as, ekwās, patrjās, spekjēs.

Stems in ja/ī produces a genitive singular in -ās; as, potnjās.

5. Dative-ablative singular in -āi, ekwāi, patrjāi.

NOTE. This LIE -i comes probably from an older PIE general dat. *-ei ending; as, *h1ekweh2-ei ekwāi. A dat.-abl. ending *-ei is also found for stems in ē and in ja/ī.

6. Locative singular in -āi; as, ekwāi, patrjāi.

7. Instrumental singular in -, -ā-bhi, -ā-mi; as, ékwābhi, pátrjābhi.

 

f. ekwā

adj. f. cowij

f. potnja

f. spekjē-

NOM.

ekwā

cowij

potnja/potnī

spekjēs

VOC.

ekw

cowij

potnja/potnī

spekjē

ACC.

ekwām

cowijm

potnjam/potnīm

spekjēm

GEN.

ekwās

cowijs

potnjās

spekjēs

ABL.

ekwād

cowijd

potnjād

spekjēd

DAT.

ekwāi

cowiji

potnjāi

spekjēi

LOC.

ekwāi

cowiji

potnjāi

spekjēi

INS.

ékbhi

cowijbhi

potnjābhi

spekjēbhi

There is only one example from this declension with a proterodynamic inflection (see §4.7), namely the word for ‘woman’: nom. cenā, gen. cnās, cf. O.Ir. nom. ben, gen. mná, Skt. nom. jánis (gn), gen. gns (jányur). Beekes (1995).

4.2.3. The Plural in the First Declension

1. The following table presents the plural paradigm of the a-stem declension.

a-Declension Plural Paradigm

NOM.-VOC.

-s

ACC.

-ns

GEN.

-m

DAT.-ABL.

-bhos (-mos)

LOC.

-su (-si)

INS.

-bhis (-mis)

NOTE. The plural is reconstructed as from PIH nom.-vocc. *-eh2(e)s, Acc. *-eh2n̥s (<**-eh2- m̥-s), gen.(-abl.) *-(e)h2om, dat.-abl. *-(e)h2bh(j)os or *-(e)h2mus, loc. *-(e)h2su, ins. *-(e)h2bhi(s); as, *h1ekweh2es ekwās. From Beekes (1995), Clackson (2007).

2. Nominative-vocative plural in -s: ekwās, patrjās, cowijs.

3. Accusative plural in -ms: ekwāns, patrjāns.

4. Genitive plural in -m: ekwām, patrjām.

5. Dative and ablative plural in -bhos, -mos, and PII -bhjas; as, ékwābhos, patrjābhos.

6. Locative plural in -su (also PGk -si); as, ékwāsu, pátrjāsu.

6. Instrumental plural in -bhis, -mis; as, ékwābhis, patrjābhis.

NOTE. The obliques have also special dialectal forms Gk. -āisi, -ais, Lat. -ais; as, Lat. rosīs<*rosais.

 

f. ek

f. cowij

f. potnja

NOM.-VOC.

eks

cowijs

potnjās

ACC.

ekwāns

cowijns

potnjāns

GEN.

ekm

cowijm

potnjm

DAT.-ABL.

ékbhos

cowijbhos

pótnjabhos

LOC.

éksu

cowijsu

pótnjasu

INS.

ékbhis

cowijbhis

pótnjabhis

ABL.

ékbhos

cowijbhos

pótnjabhos

 

4.3. Second Declension

4.3.1. Second Declension Paradigm

1.  Nouns of the second declension have a stem ending in e/o, and they are usually called thematic. They can be animates and inanimates, as well as adjectives. The inanimates have an ending -m in nom.-acc.-voc. The animates, with a nominative in -s, are generally masculine in nouns and adjectives, but there are also feminine nouns and animate (i.e. masc.-fem.) adjectives in -os, probably remains of the old indistinctness of animates.

NOTE. The o-stem declension is probably very recent in PIE – even though it happened already in PIH, before the Proto-Anatolian split – and that’s why it is homogeneous in most IE dialects. As Mallory–Adams (2007) say, “[t]he o-stems were the most productive form of declension. By this is meant that through time, especially at the end of the Proto-Indo-European period and into the early histories of the individual Indo-European languages, the o-stems appeared to proliferate and replace other stem types. In Vedic Sanskrit, for example, they constitute more than half of all nouns. High productivity is often interpreted as evidence that the o-stems are a later declensional form than many of the other stems. Highly productive forms are ultimately capable of replacing many other forms as they provide the most active model by which speakers might decline a form.”

2. The IE second declension is equivalent to the second declension in Latin (cf. Lat. dominus, dominī, or uinum, uinī), and to the omicron declension in Greek (cf. Gk. λόγος, λόγου, or δῶρον, δῶρου).

o-Declension Singular Paradigm

 

Animate

Inanimate

NOM.

-os

                                                 -om

VOC.

-e

ACC.

-om

GEN.

-os/-osjo/(-oso)/(-ī)

ABL.

-ēd/-ōd

DAT.

-ōi

LOC.

-ei/-oi

INS.

-ē/-ō

NOTE 1. This model could have been written without the initial vowel -o-; the probable origin of this vowel is the ending of some primitive ‘original’ stems in -o, while other, primitive athematic stems would have then been reinterpreted, and an -o added to their stems by means of analogy (Adrados–Bernabé–Mendoza 1995-1998). So, this paradigm could be read from a historical point of view as nom. -s, acc. -m, gen. -s, -sjo, -so, and so on.

NOTE 2. The thematic declension is usually reconstructed in the singular as from older PIH nom. *-os, voc. *-e, acc. *-om (neu. nom.-voc.-acc. *-om), gen. *-os, dat. *-ōi (<**-o-ei), loc. *-oi, ins. *-oh1, abl. *-ōd (<**-o-ed); as, dat. *wl̥kwo-ei wqōi, abl. *wl̥kwo-ed wqōd. Sometimes, the a-stem and o-stem ablative is reconstructed as from PIE *-ot or *-et, or even *-h2at (in Fortson 2004). As we have seen, -d and -t are pronounced alike at the end of the word, so the difference is mainly an etymological one.

4.3.2. Second Declension in Examples

1. Nominative singular animate in -os; as in wqos, wolf, dómūnos, lord, wīrós, man, adj. cīwós, alive.

2. Vocative singular animate in -e; as in wqe, dómūne, cī.

3. Accusative singular animate in -om; as in wqom, dómūnom, cīwóm.

4. Nominative-vocative-accusative singular inanimate in -om; as in jugóm, yoke, adj. newom, new.

5. Genitive singular in -os, -osjo, also -e/oso, -ī; as in wqosjo, jugós, dómūnosjo.

NOTE. The original genitive form -os is rare in animates, as the genitive had to be distinguished from the nominative. This disambiguation happens by alternatively lengthening the ending, as -os-jo (or -e/os-o, probably from the pronominal declension) or changing it altogether, as in Ita.-Cel. -ī. In Hittite, the genitive -os is found, so it is usually considered the oldest form, as in the athematic declension. A generalised -osjo is found in Sanskrit, Armenian, Greek and Italic, so this alternative ending must have replaced -osearly, still within the LIE community.

6. Ablative singular in -ōd, -ēd: wqōd, cīwd, jugd.

7. Dative singular in -ōi: wqōi, dómūnōi, newōi, jugṓi.

8. Locative singular in -oi, -ēi: wqoi, dómūnoi, newoi, jugói.

9. Instrumental singular in -ō, -ē: w, dómūnō, newō, jug.

 

m. wqo-

n. jugó-

adj. newo-

NOM.

wqos

jugóm

newos

VOC.

wqe

jugom

newe

ACC.

wqom

jugóm

newom

GEN.

wĺqosjo

jugós

newosjo

ABL.

wqōd

jugd

newōd

DAT.

wqōi

jugi

newōi

LOC.

wqoi

jugói

newoi

INS.

wqō

jug

newō

4.5.3. The Plural in the Second Declension

1. The thematic plural system is usually depicted as follows:

o-Declension Plural Paradigm

 

Animate

Inanimate

NOM.-VOC.

-ōs/(-oi)

-

ACC.

-ons

GEN.

-m/-ēm

DAT.-ABL.

-obhos (-omos)

LOC.

-oisu (-oisi)

INS.

-is

NOTE. The animate plural paradigm is reconstructed as PIH nom.-voc. *-ōs (<**-o-es), acc. *-ons (<**-o-m-s), gen.(-abl.) *-m (<**-o-om), dat.-abl. *-o(i)bh(j)os/-omos, loc. *-oisu (<**-o-eis-su), ins. *-is (<**-o-eis); as,*wl̥kwo-es wqōs. Inanimates have a nom.-voc.-acc in *-(e/o)h2 evolved as -ā in Sanskrit and Slavic, and -a in most dialects. A nom.-voc. (pronominal) ending -oi is also found. See Beekes (1995), Fortson (2004), Clackson (2007).

2. Nominative-vocative animate plural in -ōs; as, wqōs, dómūnōs, wīrs.

3. Accusative animate pural in -ons; as, wqons, dómūnons, cīwóns.

4. Nom.-voc.-acc. inanimate plural in -; as, jugá, cīwá.

5. Genitive plural in -ōm; as, wqōm, dómūnōm, cīwm,jugm.

6. Dative and ablative plural in -obhos, -omos; as, wqobhos, cīwóbhos.

7. Locative in -oisu, PGk. -oisi; as, wīrisu,dómūnoisu.

8. Instrumental in -is; dómūnōis, cīwis, jugṓis.

 

m. wlqo-

n. jugo-

adj. newo-

NOM.-VOC.

wqōs

jug

newōs

ACC.

wqons

jug

newons

GEN.

wqōm

jugṓm

newōm

DAT.-ABL.

wĺqobhos

jugóbhos

newobhos

LOC.

wĺqōisu

jugóisu

newōisu

INS.

wqōis

jugis

newōis

 

4.4. Third Declension

4.4.1. Third Declension Paradigm

1. Third declension nouns end in i, u (also ī, ū) and diphthong. They are found as neuter, masculine or feminine; those in ī, ū, are always feminine.

2. This declension usually corresponds to Latin nouns of the third declension in -i (cf. Lat. ciuis, ciuis, or pars, partis), and of the fourth declension in -u (cf. Lat. cornū, cornūs, or portus, portūs), and to Greek vowel stems in ι, υ, ευ, αυ, ου, ω (cf. Gk. ἰχθύς, ἰχθύος, or πόλις, πόλεως).

 

 


 

i/u-Declension Singular Paradigm

 

Animate

Inanimate

NOM.

-s

-

VOC.

-

ACC.

-m

GEN.-ABL.

-s

DAT.

-ei

LOC.

-/-i

INS.

-◌̄/-ē/-bhi (-mi)

NOTE. The i/u-stem declension is a variation of the common athematic declension of c-stems. The obliques show weak stems (root ablaut and accent shift) in some nouns.

The proterodynamic paradigm for u-stems is reconstructed in the sg. as nom. *-u-s, voc. *-eu, acc. *-u-m, gen. *-ou-s, dat. *-eu-i, loc. *-ēu, ins. *-u-h1; for i-stems nom. *-i-s, voc. *-ei, acc. *-i-m, gen. *-oi-s, dat. *-ei-i, loc. *-ēi, ins. *-i-h1. See Beekes (1995).

3. The -s can indicate nominative and genitive: the distinction is made through the full-grade or extension of the vowel before the declension, see below.

4.4.2. In i, u

1. Nominative sg. animate in -s; as, owis, sheep, noqtis, night, ghostis, guest, sūnús, son, egnis, fire, pṛtus, ford, swdhus, custom; adj. swādus, pleasant.

2. Vocative singular animate in -, or full -ei, -eu; owi, sūnéu/sūneu, swēdhu.

3. Accusative singular animate in -m; as in owim, noqtim, ghostim, sūnúm.

4. Nominative-vocative-accusative singular inanimate in -; as in mari, sea, kdi, heart, peku, cattle, deru, wood, medhu, mead, adj. swādu.

5. The genitive singular shows two inflection types:

·Type I genitive singular in -eis, -eus, also -ois, -ous; as, ghosteis, mareis, sūnéus, swēdheus, adj. swādeus.

·Type II genitive singular in -(e)jos, -(e)wos; as, owjos, noqtjos, kdejós, swdhewos, pékewos.

NOTE 1. About both types of inflection, a description was made by Wackernagel-Debrunner (Altindische Grammatik, 3 vols., 1896/1954), Kuryłowicz (The inflectional categories of Indo-European, 1964), Szemerényi (1985), etc. It is so found in Sihler (1995), Adrados–Bernabé–Mendoza (1995-1998).

They are usually said to be derived from a PIH proterodynamic inflection, originally made with a weak form (vowel change and accent shift), from which LIE simplified its root ablaut formation and accentuation; starting from the weak stems, zero grade roots were generalised and accents became static on the root or the suffix (Meier-Brügger, 2003); as, from PIH *pertus, pṛtéus, remade LIE (without root ablaut) pṛtus, pṛtewos. See below §4.7.

NOTE 2. Both types are sometimes said to be derived from two ‘original’ PIH i/u-stem accent-ablaut inflections, later merged into the known paradigms. They would have been a proterokinetic inflection, represented by **mént-ēi-s, **mént-i-m, **mt-éi-s, and an amphikinetic (?) inflection represented by **h3ew-i-s, **h3w-éi-m, **h3w-jo-s, that gave birth to the LIE types known to us. Sihler (1995): “In any case, the surviving i-stem inflections, which in this view are a sort of Chinese menu selection of items from [proterokinetic] and [amphikinetic], exhibit too much agreement in detail in InIr., Gmc., Ital. and BS to be independent innovations. Accordingly, even if this theory is accepted, the necessary leveling to get to the usual reconstruction must have been complete in the parent language.”

6. Dative in -ei, usually full -ei-ei, -eu-ei; as, ghóstejei, pékewei.

NOTE. For a dat. sg. -ei, pure stem or full ending in -i, cf. Gk. -seï (<*-t-ej-i?), O.C.S. kosti.

7. Locative in -ei, -eu, usually lengthened -ēi, -ēu, -ewi; as, noqtēi, sunu.

8. Instrumental in -ī, -ū (<*-h1), in -ē (<*-eh1) following the genitive, or  in-bhi, -mi: p, pwē.

NOTE. While the instrumental ending *-h1 (from Indo-Iranian) follows the athematic declension, the ending -mi from Balto-Slavic (hence also LIE -bhi) follows the thematic declension, but could have been a later innovation from an old trend to reinterpret athematic as thematic nouns. The older Hitt. -awet doesn’t clarify the situation.

 

Type I

Type II

 

f. ghosti-

m. sūnu-

n. mari-

f. noqti-

m. ptu-

n. peku-

NOM.

ghostis

sūnús

mari

noqtis

ptus

peku

VOC.

ghosti

sūnéu

mari

noqtei

ptu

peku

ACC.

ghostim

sūnúm

mari

noqtim

ptum

peku

G.-A.

ghosteis

sūnéus

mareis

noqtjos

ptéwos

pékewos

DAT.

ghóstejei

sūnéwei

márejei

nóqtejei

ptéwei

pékewei

LOC.

ghostēi

sūnu

marēi

noqtēi

ptu

pékewi

INS.

ghostī

sūnewē

marī

noqtī

pwē

pekū

NOTE. For information on the alternative reconstruction mari/mori, sea, see Appendix II Formal Aspects.

The Strong Type

Its inflection is similar to the consonant stems, and they have no alternating vowels before the declension; ī and ū are substituted before vowel by -ij, -uw. They are always feminine, and they cannot be inanimates nor adjectives. They are mostly PIE roots (in *-iH, *-uH), and found mainly in Indo-Iranian.

NOTE. This inflection is usually classified within the i/u-stems, for nouns where an i precedes the final laryngeal, i.e. in *-i(e)H- (It is not always clear when it comes from *-h1 and when from *-h2; they were probably completely parallel in LIE and merged). The old declension shows nom. *-iH, acc. *-ieH-m, gen. *-iH-os: in Sanskrit *-iH became generalised, showing gen. -ías, while Slavic and Germanic show -jā- < *-jəH. The *-uH stems are completely parallel with those in *-iH; cf. for ‘tongue’, a general LIE dghwā, but also found as PII dghūs. See Beekes (1995), Adrados–Bernabé–Mendoza (1995-1998).

Only NWIE forms and declension is followed here (w-, she-wolf, is apparently found in Celtic, apart from Indo-Iranian).

 

f. bhrū-

f. sū-

f. dhī-

f. wqī-

NOM.

bhrūs

sūs

dhīs

wqīs

VOC.

bhrū

sū

dhī

wqī

ACC.

bhrūm

sūm

dhīm

wĺqīm

GEN.-ABL.

bhruws

suws

dhijs

wqijs

DAT.

bhruwéi

suwéi

dhijéi

wqijéi

LOC.

bhruwí

suwí

dhijí

wqijí

INS.

bhrūbhí

sūbhí

dhībhí

wqībhí

 

 

4.4.3. In Diphthong

1. There are long diphthongs āu, ēu, ōu, ēi, which sometimes present short vowels.

NOTE. Other stems that follow this declension in the attested dialects, in ā, ē, ō, are probably remains of older diphthongs. Therefore, these can all be classified as diphthong endings, because the original stems were formed as diphthongs in the language history.

Its paradigm is reconstructed for those in -u as *-ē/ōus(s), acc. *-e/ou-m, gen. *-u-os, and for those in -i as nom. *-oi, *-is, acc. *-oi-m, gen. *-i-os. Beekes (1995). It is not a common declension, and IE nouns proper included in it are found inflected as follows: strong forms with nom. *-s, voc. *-, acc. *-m, loc. *-i; weak forms with gen.-abl. *-és, dat. *-éi, ins. *-éh1.

 

m. f. u-

m. djēu-

f. nāu-

NOM.

us

djēus

nāus

VOC.

cou

djeu

nāu

ACC.

m

djēm/dijm

nāum

GEN.-ABL.

cous

diwós

nāwós

DAT.

cowéi

diwéi

nāwéi

LOC.

cowi

djewi/diwí

nā

INS.

cow

diw

nāw

NOTE. An expected accent-ablaut reconstruction for nāu- would be strong nāu <*neh2u-, weak nau-<*nh2u- or rather *n°h2u-; however, forms in nāu- are found in Vedic and Ancient Greek dialects throughout the whole paradigm, possibly indicating older strong *noh2u- and weak *neh2u-, respectively. See Meier-Brügger (2003) for more on this question.

In zero-grade genitives there are forms with -i- or -ij-, -u- or -uw-.

NOTE 1.  Some secondary formations – especially found in Greek – are so declined, in -eus, -ewos as in Av. bāzāus, Arm.,Gk. Basileus,possibly from PIE -āus (Perpillou, 1973) but Beekes (2007) considers it Pre-Greek.

NOTE 2. Stang’s law governs the word-final sequences of a vowel + semivowel j or w +  nasal, simplified in PIE so that semivowels are dropped, with compensatory lengthening of a preceding vowel, i.e. VJM M; as, djēm, not *djewm̥; cōm, not *gwowm̥, cōns, not *gwown̥s, etc. A similar trend is found with laryngeals, *VJhm > M; as, sūm, also attested as suw, etc.

4.4.4. The Plural in the Third and Fourth Declension

1. The following table depicts the general plural system of the fourth declension.

i/u- and Consonant-Declension Plural Paradigm

 

Animate

Inanimate

NOM.-VOC.

-es

-

ACC.

-ns

GEN.

-m/-ēm

DAT.-ABL.

-bhos (-mos)

LOC.

-su (-si)

INS.

-bhis (-mis)

NOTE. An older plural paradigm for u-stems is reconstructed as nom.-voc. *-eu-es, acc. *-u-ns, gen.-abl. *-eu-om, dat. *-u-bhos, *-u-mos (<**-u-mus?), loc.* -u-su, ins. *-u-bhis, *-u-mis(<**-u-bhi?). See Beekes (1995), Fortson (2004).

2. Unlike in the singular, in which only some nominatives have an -s, in nom.-voc. plural the -s is general, and there is always one fix-grade vowel, e. So, the opposition singular-plural in -s/-es is actually /e.

3. The nom.-voc. plural animate is made in -es, in full-grade -ei-es for i, -eu-es for u, and -ijes, -uwes, for ī, ū; as ówejes, sūnewes, pŕtewes, bhruwes.

4. The accusative plural animate is in -ns: owins, sūnúns, ptuns, cōns.

5. The nom.-voc. acc. plural inanimate in -a: pekwa, marja, swādwa.

NOTE. The athematic inanimate plural ending commonly represented by -a corresponds to an older collective*-h2, which sometimes lengthened the preceding vowel (i or u) instead.

6. Gen. pl. in -om (type I usually in full -ei-om, -eu-om); as, ghóstejom, ptwom.

NOTE. The -m of the acc. sg. animate, nom.-acc.-voc. sg. inanimate and this case could sometimes be confused. It was often disambiguated with the vocalic grade of the genitive, full or lengthened, as the singular is always .

7. For the obliques plural, cf. dat.-abl. ówibhos, sūnubhos, nóqtibhos; loc. sūnusu, nóqtisu, ins. sūnubhis, ówibhis, máribhis.

 

Type I

Type II

Diphth.

 

f. ghosti-

m. sūnu-

f. noqti-

n. peku-

m. cou-

NOM.-VOC.

ghóstejes

sūnewes

nóqtejes

pekwa

cowes

ACC.

ghostins

sūnúns

noqtins

pekwa

cōns

GEN.

ghóstejom

sūnewom

noqtjom

pekwom

cowom

DAT.-ABL.

ghóstibhos

sūnubhos

nóqtibhos

pékubhos

coubhos

LOC.

ghóstisu

sūnusu

nóqtisu

pékusu

cousu

INS.

ghóstibhis

sūnubhis

nóqtibhis

pékubhis

coubhis

 

4.5. Fourth Declension

4.5.1. Fourth Declension Paradigm

1. The stem of nouns of the second declension ends in consonant or resonant, i.e. -n, -r, -s, occlusive (especially -t), and rarely -l, -m. The inflection of animates is essentially the same as that of the second or thematic declension.

2. Nouns of the fourth declension correspond to Latin nouns of first declension in -r (cf. Lat. magister, magistrī), and third declension in consonant (cf. Lat. prīnceps, prīncipis, cōnāmen, cōnāminis, etc.), and to the Ancient Greek consonant stems declension (cf. Gk. πατὴρ, πατρὸς, τάπης, τάπητος, ἡγεμών, ἡγεμόνος, etc.).

The nominative ending is -s (with occlusive, -m, -l), but there is also a nominative sg. with pure stem vowel (desinence - and lengthened ending vowel), so that the full-grade vocative is differentiated. And there is no confusion in nom./gen., as -s has a different vowel grade (nom. -s, gen. -és or -os).

Consonant-Declension Singular Paradigm

 

Occlusive, -m, -l

-r, -n, -s

NOM.

-s

- (long vowel)

ACC.

-

VOC.

-

- (full grade)

GEN.-ABL.

-es/-os

DAT.

-ei

LOC.

-i/-

INS.

-ē/-bhi (-mi)

NOTE. The so-called common, basic or athematic paradigm, the hypothetically oldest attainable PIE noun declension system, is reconstructed in the singular as nom. *-s, *-, voc. *-, acc. *-m, gen.-abl. *-(e/o)s, dat. *-ei, loc. *-i, *-, ins. *-(e)h1. See Meier-Brügger (2003), Fortson (2004). This paradigm was originally common to the i/u-stems, and it was probably inherited (and innovated) by the first and second declensions. 

Besides the usual loc. ending -i there was also the bare stem without ending. Such unmarked (‘flat’) locatives are widely encountered in modern languages (cf. Eng. next door, home), and in PIE they are well-attested in n-stems, but are rare in other consonant stems.

3. Inanimates have pure vowel stems with different vowel grades. In nouns there should be no confusion at all, as they are different words, but neuter adjectives could be mistaken in nominative or vocative animate. Distinction is thus obtained with vocalism, as in animate -ōn vs. inanimate -on, animate -ēs vs. inanimate -es (neuter nouns in -s show -os).


 

4.5.2. In Occlusive, m, l

1. Nominative sg. animate in -s; as, pods, foot, regs, king, preks, plea, bhurghs, watch-tower, ghjems, winter, nepēts, grandson, adj. bélowents, strong.

NOTE. The nom. of some stems are often reconstructed with a long vowel; as, *pōds, *rēgs, *prēks. Such forms are found in the different languages showing ablaut with lengthened grade; as, ō/o in Goth. fotus, Gk. pód-a, and ē/e in Lat. pēs, pedis. It is usually interpreted that these are levelled forms from an original o/e opposition, so the long vowel vs. short vowel becomes unnecessary for the parent language. Also, sometimes it is doubted whether the original nominative had an s, cf. *pōd(s) in Beekes (1995), because all attested languages show a lengthened vowel with either the final occlusive or -s, but not with both (cf. Skt. pt, which could derive from *ō, , or *o, Dor. Gk. pōs, Lat. pēs), what suggests a compensatory lengthening with the loss of a final consonant cluster, that was reinterpreted as the original stem in declension (e.g. to form the accusative in some languages), i.e pods *pōs *pōds (Sihler 1995).

2. Accusative singular animate in -; as, pod, reg, bhurgh, ghjem, népēt, bélowent.

NOTE. Forms in m make the accusative by lengthening the root vowel, *Vmm > *m, as a consequence of Stang’s Law, v.s. §4.4.2; as, nom. *doms, ‘house’, acc. *dōm (<*dom-), cf. Arm. tun or Gk. δῶ, or nom. *dhghōm, ‘earth’, acc. *dhghōm (<*dhghom-ṃ), cf. Skt. kām. Root nouns like these ones are quite old in the language history, and are therefore rare in LIE, which had replaced them for newer derived nouns; as, domos, house, or dhghós, earth.

3. Vocative singular animate in -; as, pod, reg, bhurgh, bélowent.

4. The nom.-voc.-acc. singular inanimate in -; as sal, salt, part. bheront.

5. Genitive singular in -os, -es; as, pedós, rēgos, bhurghos, ghimós, salós, népotos, bélowentos, bhérontos.

NOTE. Older root nouns made the genitive-ablative often in -s, -é/ós, i.e. ablaut and accent on stem vowel; as, for strong nom. *doms, weak gen. *dems or *dmés/dmós, for strong nom. *dhghōm, gen. *dhghmés/*dhghmós or (possibly already in the proto-language) metathesised *ghdhmés/*ghdhmós. They appear in IEDs mainly frozen in compounds (cf. dems- in demspots), because most of them were reinterpreted. On forms like *dmés or *pdós, “[u]nsurprisingly, such forms have been largely eliminated from the attested paradigms.” Sihler (1995). For the original lengthened grade rēgos/rēges, see §4.7.

6. Dative singular in -ei: pedei, rēgei, bhurghei, bhérontei.

7. Locative singular in -i: podi, regi, bhurghi,bélowenti.

 

m. pod-

f. prek-

f. bhurgh-

n. bheront-

NOM.

pods

preks

bhurghs

bheront

VOC.

pod

prek

bhurgh

bheront

ACC.

pod

prek

bhurgh

bheront

GEN.-ABL.

pedós

prēkos

bhurghos

bhérontos

DAT.

pedéi

prēkei

bhurghei

bhérontei

LOC.

pedí

preki

bhurghi

bhéronti

INS.

pedbhí

prēkbhi

bhurghmi

bhérontbhi

4.5.3. In r, n, s

1. Nominative singular animate in - with lengthened vowel; as in mātr (also mtēr), mother, elōr, swan, kwōn, dog, osēn, autumn, ōs, mouth, mōs, character, adj. juwōn, young, mātōr, motherless.

NOTE. The lengthening of the predeclensional vowel in stems in r, n and s stems has been explained (Szemerényi’s Law) as a consequence of an older (regular PIE) nom. -s ending; as, **ph2tér-s→*ph2tr, **kwon-s→*kwōn, etc.

Common s-stems without lengthening include neuter stems (which are not usually marked in the nominative); as, opos, work, nebhos, cloud, spes, hope, etc.

Adjectives usually end in -es; as, sugens, well-born, of good stock (cf. Gk. eugens, O.Ind. sujanā).

2. Accusative sg. animate in -m; as in māter, élor, kwon, ópos, júwon.

3. Voc. sg. animate in - with full vowel; as mātér, élor, kúon, juwon, opos, sugenés.

4. The nom.-acc.-voc. singular inanimate in -; as in nōm, name, genos, kin.

The adjectives in -s have a neuter in -es; as, sugenés.

5. Genitive singular in -os, usually with an e, not an o, as the final stem vowel; as, éleros, nómenos, bhugenos, júwenos, nébhesos, génesos, ópesos, ēsos, spēsos; but cf. zero-grade in old stems, asmātrós, kunós, and also o, as mosós.

NOTE. Athematic nouns made the genitive in -es, -os; even though some reconstruct only one gen. ending for an ‘original’ paradigm, we prefer to write always -ós for pragmatic reasons; as, kunós instead of kunés.  Examples of an apparently old, so-called static inflection, is found in PII nom. -tr, gen. -tr-s,acc. pl. -tr-ns; remains of this rare paradigm are also found in Gmc. dialects forfather’ in gen. patŕs.

6. Dative singular in -ei, mātréi, élerei, kunei, júwenei, ópesei, sugenesei.

7. Locative singular in -i: māteri, éleri, kuni, júweni, ópesi, sugenesi.

8. Instrumental singular in -ē or -bhi/-mi: mātbhí, élerbhi, patṛbhí, kunbhí, júwenbhi, ópesbhi, sugenesbhi.

 

m. kwon-

f. pater-

n. genes-

n. nom-

adj. sugenes-

NOM.

kwōn

patr

genos

nōm

sugens

VOC.

kwon

patér

genos

nōm

sugenés

ACC.

kwon

pater

genos

nōm

sugenes

G.-A.

kunós

patrós

génesos

nómenos

sugeneses

DAT.

kunéi

patréi

génesei

nómenei

sugenesei

LOC.

kuní

pateri

génesi

nómēn

sugenesi

INS.

kwbhí

patbhí

génesbhi

nómenbhi

sugenesbhi

NOTE. “Where the derivation is transparent, neut. s-stem nouns were built to the tonic e-grades of verb roots. The stem had the form *-os in the (endingless) nom./acc.sg., *-ōs<**-osH2 in the nom./acc.pl; the stem of the remaining cases and numbers was an invariant *-es-”. Sihler (1995), who further reconstructs the inflected forms of genos, with root accent throughout the paradigm.

4.5.4. The Plural in the Fourth Declension

With a paradigm common to the third declension, here are some inflected examples.

 

m. pod-

f. prek-

m. kwon-

f. māter-

n. genes-

N.-V.

podes

prekes

kwones

māteres

génesa

ACC.

pods

preks

kwons

māterṇs

génesa

GEN.

pedóm

prēkom

kunóm

mātróm

génesom

D.-A.

pedbhós

prēkbhos

kwbhós

mātbhós

génesbhos

LOC.

pedsó

prēksu

kw

māt

génesu

INS.

pedbhí

prēkbhi

kwbhí

mātbhís

génesbhis

NOTE. The plural of adj. mātōr, motherless, consists of masc./fem. nom. mtores, neu. nom-voc.-acc. mtora gen. matróm, etc.

4.6. Variable Nouns

4.6.1. Many nouns vary in declension, and they are called heteroclites.

4.6.2. Heteroclitic forms are isolated and archaic, given only in inanimates, as remains of an older system, well attested in Anatolian.

4.6.3. They consist of one form to mark the nom.-acc.-voc, and another for the obliques, usually r/(e)n; as, bhem/bhémenos, thigh,  ghēs/ghésenos, hand, gut/gútenos, throat, kow/kówenos, cavern, ūdh/ū́dhenos, udder, wedh/dhenos, weapon, etc. 

4.6.4. Different paradigms are also attested:

·  Opposition r/(e)n- (lengthened ending); as, jeq/jeqóneros, liver,

NOTE. For PIE jeq, cf. Ved. yákt, Gk. hēpar, Lat. iecur, Av. yākarə, and compare its Obl. Skt. yakn-ás, Gk. hpat-os<*hēpn̥t-, Lat. iecinoris.

·  Alternating with other suffixes; as, gheim/gheims/ghjems, winter, skīw/kīwōn/skinōn, shinbone, later column, wēs/wēstós, spring, wed/wédenos/wodā, water, swep/swopnos, dream.

·  Formed from the consonant r or n of the heteroclite; as, pāw/pūr/puōn, fire, nom/nómeros/nómenos, precision, number, Gk. skōr (gen. skatos), Hitt. šakkar (gen. šaknaš), Lat. -scerda, “shit”.

4.6.5. The heteroclites follow the form of the genitive singular when forming the obliques. That is so in accent-ablaut and in the lengthening before declension.

4.7. Inflection Types

1. While in o-stems and generally in a-stems there are generally no accent-ablaut changes, within the c-stems it is possible to distinguish different old accent-ablaut patterns, which make this a more complicated declension.

NOTE. Fortson (2004): “To understand athematic nomina inflection, one must distinguish between the so-called strong and weak cases. The strong cases differ from the weak cases typically in where the accent is located and which morpheme is in the full grade; most commonly, the full grade and the accent shift rightward in the weak cases, comparable to the shift seen in most athematic verbs.”

Fortunately it is the least productive of all declensions (i.e. it is infrequent in new nouns), and most stems that included accent-ablaut patterns were progressively substituted for other stems in LIE.

NOTE. Fortson (2004): “All athematic nouns consisted of three parts: root, suffix, and ending. (...) As already alluded to, these three morphemes could each show up in different ablaut grades depending principally on the position of the accent, which could fall on any of the three. According to the standard theory, in any given case-form of an athematic noun the unstressed morphemes appeared in the zero-grade, while the stressed morphemes were in a grade “stronger” than zero-grade- that is, one with a vowel, generally e, but also o.”

2. We can distinguish at least two kinds of inflections in PIE, which differ in ablaut and accent. These types are called hysterodynamic (dynamis ‘accent’; hystero- ‘more toward the back’) and proterodynamic (protero- ‘more toward the front). In these types the accent shifts rightward in the weak cases from its position in the strong cases (Beekes 1995).

NOTE. Other inflection pattens are apparently inherited from the parent language – the definition, pattern and the very existence of some of them is still debated –, but they are usually not found in IEDs, if not in frozen remains of the older system. Therefore, it is not interesting to complicate the regular athematic system further, and we will treat such remains as irregularities of a common LIE system. For more information, see below.

3. The neuters followed the proterodynamic inflection, and do not have nom. -s, acc. -m. Some of the masculine-feminine nouns follow the proterodynamic inflection too.

4. In hysterodynamic nouns, the suffix is accented in the strong cases, the ending in the weak. So e.g. in the word for ‘father’, we have strong nom. sg. pa-tr s, acc. sg. patér-, but weak gen. pa-tr-és.

NOTE. Fortson (2004): “Root nouns with mobile accent (on the root in the strong cases, on the endings in the weak cases) are often considered as belonging to this type, though the fit is not exact: they have full grade or lengthened grade of the root in the strong cases and full grade of the ending in the weak cases”. For pragmatic purposes, we will include nouns formed with only a root and ending among hysterodynamics, too. The most common type has o-grade of the root in strong cases, and e-grade in the weak cases. So e.g. for foot, nom. pods, acc. pedós.

In the so-called amphikinetic nouns, the root is also accented in the strong cases, the ending in the weak, and the suffix is typically in the lengthened o-grade (rather than the expected zero-grade) in the nominative singular, and ordinary o-grade in the accusative singular.  The old word for dawn (found in PGk and PII) belongs here; áus-ōs (c. Gk. aúōs), acc. sg. áus-os-m̥ (Ved. uṣasam ), gen. us-s-és (Ved. usás).

5. In proterodynamic nouns, the root is in the full grade and accented in the strong cases, and both accent and full grade shift to the suffix in the weak cases.

NOTE. According to Fortson (2004): “Most i- and u-stems in Sanskrit appear to have been proterokinetic, such as Ved. nomin. matís ‘thought’, accus. matím, genit. mates, from PIE *mén-ti-s, *mén-ti-m, *mn̥-téi-s.” Vedic had generalised the zero-grade of the root throughout the paradigm, as the other languages; hence LIE mtis, mtéis. The same could be said of the remodelling of old *pertus, *pṛtéus, remade pṛtus, pṛtewos.

As we have seen, some i/u-stems behave in parallel to those athematic nouns, showing zero-grade ending -i-, -u- in strong cases (as nom. -i-s, -u-s, acc. -i-m, -u-m, and full-grade -ei-, -eu- in weak cases (nom. -ei-s, -ei-os, -eu-s, -eu-os). By the time of LIE these nouns do not usually show ablaut in their root syllable. Common examples are found of verbal abstract nouns in -ti-, one of the most common groups of i-stems; also common were abstract nouns in -tu-, although this usually appears in post-LIE infinitives.

Remodellings like these led to a new class of proterodynamics that descriptively had o-grade of the root in the strong cases and zero-grade in the weak. Among animate nouns of this type, the best attested is the word for ‘dog’, nom. kwōn, acc. kwōn-, gen. kun-ós (or kun-és).

Excursus: Nominal Accent-Ablaut Patterns

While nominal accent-ablaut patterns are very interesting for internal reconstruction, they remain a marginal issue for the largely thematic Late Indo-European language, and more so for IEDs. The state of the art about such patterns is summed up in a recent article by Alvin Kloekhorst <http://www.kloekhorst.nl/KloekhorstIENominalAblautPatterns.pdf>.

The author exposes results based on testimonies of Hittite, in which he is a renowned expert. Accent-ablaut paradigms can be reconstructed according to two systems, Leiden and Erlangen, which correspond to two different times, which we could roughly assign to PIH and (early) LIE respectively. He explains the transformations from one system to the other, from which a system of two models (static and dynamic) turns into a system of four (static, proterodynamic, hysterodynamic, and resulting from analogical changes, amphidynamic).

The system of four models of Erlangen is explained by Frazier in the following link, with examples in p. 112: <http://roa.rutgers.edu/files/819-0406/819-FRAZIER-0-0.PDF>  and summed up in <http://www.unc.edu/~melfraz/ling/frazier-UCLA-handout.pdf>.

The Erlangen system is therefore nearer to the Late Indo-European system, because the reconstructed patterns are more recent. However, in some cases the reconstructed forms – since they look for a logical agreement with Anatolian – are clearly different from non-Anatolian results; as e.g. the oblique cases of *nebhos – see note 34 in Kloekhorst’s article and Clackson (2007, p. 94).

About root nouns of the type Lat. rex, lux, pēs, etc., the problem gets more complicated. A priori there are only two possibilities about the distribution of the tone: either static (with columnar accent on the root), or kinetic/dynamic, with accent on the root in the strong forms, and on the ending in the weak forms. However, to make grades and vowel ablaut fit into common patterns is more difficult (see Appendix III.1 for some examples).

Kloekhorst shows some problems in the ablaut models of root nouns. Examples are:

·  The word for ‘foot’, which is usually reconstructed as having a static paradigm, *pd-s, *pód-m, *péd-s, seems to be mobile in Hittite: acc.pl. pāduš < *pód-ms vs. gen.pl. patān < *pd-óm

·  Av. vāxs shows gen. vacō, but Old Indian and Latin show systematically lengthened grade (gen. O.Ind. vācáh, Lat. uōcis); for example Brugmann’s Law, if applicable, should apply to O.Ind. and Av. See <http://www.fas.harvard.edu/~iranian/Avesta/a14_lesson11.pdf>, p.4 for Avestan grammar. Hence it could be that O.Ind. and Av. alternate with different ablauts.

·  The inflection of *gwow is static in Old Indian, but dynamic in Greek.

·  Inflection of *ker(d) is dynamic in O.Ind., but is the only example of static inflection in Greek, so this could correspond to an older model.

·  The position of the accent in weak cases of *kwon are not coincident in Gk. and O.Ind.

·  *suH and *dieH show alternating forms in O.Ind.

·  The reconstruction of *j(e)uH needs *juHs- in Sanskrit and Lithuanian, but *jeuHs- in Slavic (de Vaan).

·  Also, Greek has lost the ablaut changes in those monosyllabics (e.g. χείρ); Latin has merged diphthongs and has lost the original ablaut patterns; O.Ind. has merged vocalic ablaut *e/o into a; in Hittite, root nouns are limited to some words in transition to other inflection types (pat, gwau, siw-at, nekuz, karaz, tekan, kess-ar, gim(a)-, happar-, tuekk-). Frequently root nouns of one language are only found as derived nouns in others, and because of that coincidences are scarce.

Paul Kiparsky helps clarify the problem with the following recent article <http://www.stanford.edu/~kiparsky/Papers/ucla_IE_09.submitted.new.pdf>, although in p.7 he exposes the difficulties of fitting the attested results with the created models. In section d) p. 12 of this article, Kiparsky resorts to the artifice of ‘accented morphemes’ to explain the brief *o of O.Ind. dat. gáve. It is possibly derived from weak form *gwew.

Clackson (2007) proposes a different alternating paradigm; in p.86 he reconstructs *pods, gen. *peds., while Kloekhorst in the note of p. 3 proposes *pōds/peds. In Greek and Old Indian, these nouns aren’t acrostatic.

About Lat. rex, in Latin, Celtic and Old Indian the declension of this word shows systematically lengthened grade – in Latin letter x is considered as two consonants, so the syllable is in every case long. For results in Celtic, see p. 19 in the article <http://www.univie.ac.at/indogermanistik/download/Stifter/oldcelt2008_1_general.pdf>.

Fort he reconstruction of this root noun there are, a priori, at least four possibilities:

a) Nom. sg. *Hregs / gen. *Hrgós, as Gk.  nom. ἀνήρ <*Hner-s gen. ἀνδρός<*Hnrós. It seems that the Vedic (Sharfe) Celtic (McCone) words for king and queen point to that form.

b)  Nom. sg. *Hregs / gen. *rēgos, as Gk. κῆρ, where the weak form is not *Hr̥g, but *reHg-.

NOTE. It has been assumed that it is acrostatic, with alternating full/long, based on:

·  The attested nominative could be analogical in relation to the weak forms.

·  The long vowel appears systematically in the attested languages.

·  The derived form in O.Ind. rājan has lengthened grade, when long grade was expected (there could be a special form in Gk. arēgn). Cf. Av. bərəzi-rāz with Osc. meddiss (zero-grade).

c) Nom sg. *rēgs, gen. *rēgós, taking as model O.Ind. vac (if the long vowel is not derived from Brugmann’s Law), or náuh.

d)  Nom sg. *rēgs, gen. *regós, taking as model Av. vac, or O.Ind. ap. In this case we should have to explain the difference in results within Old Indian as a result of later analogies with nau.

It is therefore quite difficult to obtain unequivocal reconstructions for these old athematic root nouns, so the forms reconstructed are usually just one possibility of the alternating patterns.

Unfortunately it often happens that we can demonstrate one form and the opposite at the same time. Fernando López-Menchero has followed these criteria in the reconstruction of the LIE lexicon in Appendix II, according to the available data on LIE dialects:

 

Acrostatic:

·  *gwow-, Lat. Bōs o/e, maybe o/o

·  *knouk-, Lat. Nux o/e

·  *pod-, Lat. Pēs  o/e

·  *woq-, Lat. Vox o/e  AK maybe dynamic of the type ō/o – ō/ō (a.i.) or maybe ō/e (Avestan), or o/o (Greek)

Dynamic:

·  *djeHw-, Lat. Diēs e/

·  *(H)reHj-, Lat. Rēs e/

·  *weik-, Lat. Vīcus  e/

·  *leuk-, Lat. Lux  e/

·  *deuk-, Lat. Dūx e/

·  *sneich-, Lat. Nix e/

·  *ghjems-, Lat. Hiems e/

·  *weis-, Lat. Vīs e/ (also *wejos/es)

Special dynamics:

·  *āp/ap -,  ā/a

·  *kwon-, Lat. Canis  o/

·  *naHw-, Lat. Nāuis  ā/ā

·  *dheghom-, Lat. Humus  e-o/-

·  *mōs/mosós, Lat. Mōs  ō/o (or maybe ō/ō?)

Without reference – these nouns are supposedly full/lengthened within an acrostatic system, hence they are all put into the same group (even though Lat. prex is not inflected with a long vowel):

·  *leg- Lat. Lex

·  *reg- Lat. Rex

·  *prek- Lat. Prex

·  *pag- Lat. Pax

·  *H3ops (ēpos) – the laryngeal does not colour the following vowel (Eichner’s Law)

·  *rosrēsos, Lat. Rōs (H3ops and ros could be inflected like mōs and vice versa)

·  *jeus could be inflected on the basis of *jūs; on the other hand, it is commonly reconstructed meus as *mūs<*muHs, and *sūs<*suHs, so the weak forms would have a long *ū (Kiparsky in p. 16 shows the same paradigm with *séuh-nu-s,  *suh--s).

Some stems that only appear dialectally as root nouns have been transferred to their general o-stem or i-stem declension:

*kerd- ‘heart’, *mems- ‘meat’, *meHns- ‘month’, *nas nose’, *ner ‘man’, *noqt- ‘night’; stems in *os/es, *puwos ‘pus, rottenness’ and *bhāos ‘light’ and ‘speech’. *sal, *dont (assimilated to a root noun) and *weis ‘strength’ are reconstructed as alternating.

gwen- has been transferred to the feminine declension. For its consideration as root noun, see Jasanoff <http://www.people.fas.harvard.edu/~jasanoff/pdf/Old%20Irish%20be_.pdf>.

4.8. Number Developments: The Dual

4.7.1. While singular and plural are relatively fixed values, the dual has proven to be unstable; it is found in Ind.-Ira., Gk., BSl. and Cel.

NOTE. Generally speaking, the rise and decline of the dual may be directly investigated in individual IE languages, e.g. in Greek, in which the dual is a fixed component of the language, while it is missing altogether in Ionic and Lesbian. The origins of the dual might be found in two word types: the personal pronoun and terms for paired body parts (as ‘ears’, ‘eyes’, ‘breasts’, etc.). It is uncertain whether the dual was an old category that gradually disappeared, or more likely a recent (Late Indo-European) development that didn’t reach all IE dialects. See Meier-Brügger (2003).

4.7.2. The formations vary depending on the stems.

1. The nominative-accusative-vocative is made:

·  Stems in a: in -āi for ā; in -ī for ja/ī.

·  Stems in o: Animates in -ōu (alternating -ō/-ōu); inanimates in -oi.

·  Stems in i, u: Animates and inanimates in -ī, -ū.

·  Consonant stems: in -e (not general).

NOTE. The endings are usually summed up as a common PIE *-h1(e), *-(i)h1. See Fortson (2004).

2. The obliques were still less generalised, the system being reconstructed as follows:

·  Genitive in  -ous,

·  Dative-ablative in -bhos/-mos,

·  Locative in -ou,

·  Instrumental in -bhis/-mis.