Appendix:Proto-Indo-European declension

English Wikipedia has an article on:

Proto-Indo-European noun

Wikipedia

This article is not a synchronic overview of Proto-Indo-European (PIE) nominal inflection, but a comparative overview of reflexes of the reconstructed PIE paradigms in attested daughter languages of major branches.

PIE nominal words had the morphological categories of gender, number and case. The category of number was shared with verbs, while gender and case were not.

There were three numbers: singular, dual and plural. There were three genders: masculine feminine and neuter. There were eight cases: nominative, accusative, vocative, genitive, ablative, dative, locative and instrumental case.

Nominal stems, as well as verbal stems, are divided into two formal classes, thematic and athematic, the former ending in the thematic vowel. Their inflection partially differed. The following is the table of the basic inflectional endings:

Some properties of PIE nominal inflection had considerable impact on the development of declensional paradigms in the daughter languages. Syncretism was one of them:

• no declension class had more than 3 different forms in dual
• neuter gender always had identical nominative, accusative and vocative forms in all three numbers
• ablative differed from genitive only in thematic stems
• in plural, ablative and dative are always identical, as well as nominative and vocative

Zero-endings in the non-neuter nominative singular chiefly arose by Szemerényi's law. Stems in -h₂, regularly feminine gender, lacked an overt nom. sg. ending for unclear reason.

PIE had several declensional patterns, fundamentally divided into mutable and fixed stems: the former changed the root and/or suffix vocalism throughout the paradigm, while the latter exhibited no such alternation. Generally, immutable stems are those on *-o- (i.e. the thematic stems), and the stems on *-eh₂, *-uh₂ and *ih₂.

Thematic stems or o-stems are the most important and widespread types of nouns. Paradigm is given for the masculine noun *wĺ̥kʷos (“wolf”) which was preserved in most of the daughters.

1. nominative singular: The ending *-os arose from the thematic vowel *-o- and the nominative singular ending *-s. It was preserved in Ancient Greek, and Old Latin (changing to Classical Latin -us). In Sanskrit, PIE */o/ changed regularly to Proto-Indo-Iranian */a/. In OCS -ъ is a regular reflex of word-final PIE *-os: *-os > Proto-Slavic -ŭs > Common Slavic *-ъ. Gothic -s is a reflex of PIE *-os -> -oz -> Proto-Germanic --az, with devoicing of /z/ and regular loss of unstressed word-final short vowels, not including diphthongs.
2. vocative singular: PIE *-e is not a real ending per se, but an e-grade of the thematic stem with zero-ending *-Ø. In Sanskrit, PIE */e/ changed regularly to Proto-Indo-Iranian */a/. Word final unstressed /e/, /a/ and /o/ was regularly lost in the development stages of Proto-Germanic and thus it is not preserved in Gothic.
3. accusative singular: The ending *-o-m arose from the thematic vowel *-o- and the accusative singular ending *-m. It was preserved as Old Latin -om yielding later Classical Latin -um. In Greek word-final PIE *-m regularly changed to -ν. OCS reflex of -ъ is regular: PIE *-om > *-um > *-un > Proto-Slavic *-u > Common Slavic -ъ. In Sanskrit, PIE */o/ changed regularly to Proto-Indo-Iranian */a/.
4. instrumental singular: The ending *-oh₁ has been preserved in Lithuanian and Vedic Sanskrit (Classical Sanskrit ending -eṇa is taken from pronominal paradigm; cf. téna, instrumental singular of तद् (tád, “it; that”)). In Latin, instrumental has undergone syntactic merger with the ablative. In Slavic the original ending has been preserved in some fossilized expressions such as Russian вчера (včerá, “yesterday”), while the OCS ending -mь has been taken from the athematic paradigm (on which see below).

   Sometimes this ending is alternatively reconstructed as -eh₁ due to the ambiguity of Indo-Iranian evidence (where Proto-Indo-Iranian */a/ could have resulted from both PIE */e/ and */o/, who merged to PII */a/), but numerous auxiliary evidence points to *-oh₁.

5. dative singular: PIE ending *-ōy resulted from the contraction of the thematic vowel *-o- and the dative singular ending *-ey. In Greek it has been preserved as -ωι, with final -ι eventually lost in the historical period yielding ῳ with iota subscript (all long dipthongs lost the final glide in the same way). Similarly in Latin: Proto-Italic *ōy was preserved in Osco-Umbrian (in Oscan as -úí), yielding eventually Latin -ō. Old Latin has some instances of the older ending: DVENOI in Duenos Inscription and NVMASIOI in Praeneste fibula. In Sanskrit the expected ending -āy has some kind of a particle (or postposition) *-a appended. Standard Aukštaitian Lithuanian -ui is by analogy to ā-stems (see below), while the Žemaitian dialects have the endings -ū, -ō, which regularly correspond to OCS & Common Slavic ending -u, and can be derived from Proto-Balto-Slavic *-ōw, which could have regularly reflected PIE *-ōy.
6. ablative singular: PIE ending *-ōd or *-ōt resulting from the contraction of the thematic vowel (*-e- or *-o-) and the ablative singular ending *-od or *-ot. It has been preserved in Old Latin as -ōd (yielding Classical Latin -ō) and Sanskrit -āt. In Balto-Slavic PIE ablative has taken over the function of genitive, and OCS and Lithuanian genitive singular endings -o and -a respectively can be traced as regular reflexes of PIE -*ōd (> Balto-Slavic *-a). In Greek the ending survived in Doric adverbs of place-from-which such as ὄπω (ópō, “whence”), with the similar adverbial use as in Latin prīmō (“at first”) or tūtō (“safely; lit 'in one piece'”).
7. genitive singular: The oldest ending of PIE genitive singular is the same as nominative singular -os, and that state of affairs has been preserved in Anatolian, e.g. Hittite 𒀜𒋫𒀸 (attas, “father”), both nominative and genitive singular. On the other hand, the genitival adjective suffixes of Hittite -ašša-, Luwian -ašša/i- and Lycian -ahe/i- are all regularly derivable from PIE *-osyo, which itself, taken from originally adjectival paradigm, became the predominant genitive singular suffix of thematic nouns in the post-Anatolian period of PIE. The suffix *-osyo can be explained as the basic form of *os with suffixed relative pronoun *-yo (cf. Sanskrit यद् (yás, “who, which”)), and is attested as expected in Sanskrit -asya and Homeric Greek -οιο (later by loss of -ι- yielding first -οο and then -ου or -ω, depending on the dialect). It has been also preserved in some Old Latin inscriptions such as Lapis Satricanus (POPLIOSIO VALESIOSIO SVODALES "the accustomed associates of Publius Valerius"), while the Classical Latin ending -ī is by origin most probably a possessive adjective. Balto-Slavic genitive case ending is by origin PIE ablative (see above).
8. locative singular The ending is reconstructed as either *-ey or -*oy. In Greek both have been preserved in adverbs such as οἴκοι (oíkoi, “at home”) and ἐκεῖ (ekeî, “in that place, there”). Similary to Greek, Latin has preserved traces of locative case only in adverbs such as domī (“at home”) and humī (“on the ground”), but the Italic branch is ambiguous on which of the two PIE endings is the exact source. Sanskrit ending -e could reflect both PIE *-ey and *-oy (merging to Proto-Indo-Iranian diphthong *-ai), while OCS -ě could reflect only PIE *-oy (> Proto-Balto-Slavic diphthong *-aj, regularly reflected as Common Slavic jat *-ě). It appears that the original ending was *-ey later being ousted by *-oy, itself resulting by suffixation of the locative singular suffix *-i onto the thematic vowel *-o-.
9. dual - Dual cases are very difficult to reconstruct formally. The reconstructable ending *-oh₁ in NAV might have been originally identical the ending of instrumental singular; the meaning of dual could've evolved from instrumental (i.e. comitative) sense, i.e. in the syntagm such as "wolf with wolf" the first element could have been dropped, and the second element ("with wolf") reanalyzed as dual ("two wolves"). At any case, PIE *-oh₁ > post-PIE *ō which regularly yielded Greek -ω, Vedic Sanskrit -ā (Classical Sanskrit -au was later Indic innovation in the historical period) and OCS -a. In Latin dual is preserved in some isolated forms such as ambō (“both”). Other dual cases are difficult to reconstruct.
10. nominative plural: PIE ending *-ōs is result of the contraction of the thematic vowel *-o- end the nominative plural ending *-es. It has been preserved only in Sanskrit as -ās. Balto-Slavic, Greek and Latin have taken then ending *-i from the pronominal paradigm (cf. Sanskrit te < PIE *to-y, nominative plural of sás "who, he"). Classical Latin suffix -ī reflects Old Latin -oe and more common -ei.
11. accusative plural: PIE ending *-ons has been regularly reflected in all the languages.
12. instrumental plural: PIE ending *-ōys results from the contraction of the thematic vowel -o- and the instrumental case ending *-oys. It has been regularly reflected in all the languages
13. dative / ablative plural: the reconstruction is not certain. Latin -īs derives from Old Latin -eis, which derives from instrumental plural ending -ōys, just as Greek -οις, though via a different route. Balto-Slavic and Germanic have endings with the -m- element: In Balto-Slavic the ending was *-mus which was preserved in Old Lithuanian (modern Lithuanian -ms is by syncope). Alternative reconstructions of the ending as *-obʰos is primarily based on Sanskrit and Latin. According to Beekes 1995, the ending *-mos was originally ablative plural marker, while the ending -bʰos was dative plural marker, and these two cases were eventually syncretized but with different outputs in the daughters.
14. genitive plural: The ending *ōm resulting from the thematic vowel -o- onto which the genitive plural marker *-om was suffixed. Greek reflex -ων is regular, with the word-final change of -m to -n. Old Latin has -om, and Classical Latin -um, but which remained productive only in two obscure classes of nouns: words construing with mass nouns (e.g. passum "of paces", sēstrium, archaic gen. pl. of sēstrius, later reinterpreted as neuter singular) and certain stereotyped expressions (duumvirum " of two men", praefectus fabrum "commander of the workers"), as well as some isolated words (socium "of allies", deum "of the gods" etc.). The ending -ōrum was modeled after genitive plural of feminines -ārum, itself imported from the pronominal inflection.
15. locative plural: The ending *-oysu resulted from the base thematic vowel '-o-', onto which the locative case ending *-isu was suffixed, itself formed by agglutination of the plural ending *-su onto the locative singular ending -i. All the surviving reflexes in the daughter languages are regular.

O-stem neuters differ from the o-stem masculines only in NAV. Lithuanian has lost neuter gender so it is omitted from the table. The paradigm is given for the neuter noun *yugóm (“yoke”).

• Most of the reflexes are by regular sound changes. In OCS the nominative ending -o cannot be directly from PIE -om (which would regularly yield OCS -ъ, cf. accusative singular ending above), so it's generally held to originate from pronominal ending in NAV neuter gender *-od (cf. Sanskrit तद् (tád, “that”)) which has been analogically transferred to nouns.
• In Ancient Greek the nominative ending -ω is of uncertain origin, probably taken from the masculine nominative dual ending.
• In Sanskrit plural form, on the original pre-form *yugā the element -n- has been added analogically to n-stems (see below), and the ending -i < *h₂ (the regular reflex of syllabic laryngeal in Sanskrit) for NAV of neuter-gender athematic nouns (see below).
• The plural ending of neuter gender formed with the suffix *-eh₂ was in PIE a collective noun which agreed with the verb in singular. That state of affairs has been preserved in Hittite and Ancient Greek, e.g. πάντα ῥεῖ "all things flow", where πᾶς (pâs, “every, each”) is in nominative plural, and ῥέω (rhéō, “I flow”) is in third-person singular.

PIE eh₂-stems or ā-stems (the traditional name) were mostly of feminine gender. Some daughter languages have masculine-gender exceptions (e.g. OCS слоуга (sluga, “servant”), Lithuanian el̃geta (“beggar, poor man”), Latin verna (“homeborn slave”)). Sometimes this inflectional type is called senā-type, after Sanskrit सेना (sénā), which also belongs to this type.

This type encompasses Greek and Latin "first declension". Of five subtypes of Greek first declension, the subtype with nominative in -ᾰ (e.g. θάλασσα (thálassa, “sea”)) is historically unrelated. The other four types all evolved from this inflectional type in PIE.

Example paradigm is given for *h₁éḱweh₂ "mare", eh₂-stem derivation of *h₁éḱwos (“horse, stallion”). For Greek, which has no such eh₂-stem derivation of the inherited word ἵππος (híppos), the inflection is provided for the word ἰδέᾱ (idéā, “look, appearance”), and in OCS for the word жєна (žena, “woman”).

1. nominative singular: There is no ending, the bare eh₂-stem is used. In Greek it regularly yields -ᾱ or -η, depending on the dialect and the preceding vowel (in Attic, ᾱ changing to η except after ε, ι and ρ). The ending -ς in Greek masculine ā-stems (e.g. νεᾱνίᾱς (neāníās, “young man”) is of secondary origin. In Latin pre-form *-ā has yielded -a due to iambic shortening in disyllabic nouns, and then analogically spreading to all the other nouns of this type.
2. vocative singular: The ending is the same as nominative singular, except that the laryngeal drops without lengthening the preceding vowel, usually explained as regular loss of a laryngeal in pause, i.e. at the absolute end of words and sentences. PIE vocatives were phonologically and syntactically separate from the clauses they appeared in, with the result of laryngeal being dropped, leaving only the short vowel. This explains the short vowel in Ancient Greek vocatives such as νύμφᾰ (númpha, “Oh maid!”) and OCS -o (regularly from post-PIE *a, PIE *ā would've yielded OCS a). The rest of Greek feminines, as well as all the Latin, use the nominative singular form as vocative.
3. accusative singular: PIE *-eh₂m (with non-syllabic *-m) > post-PIE *ām, regularly yielding Greek -ᾱν, -ην (with regular Greek change of word-final -μ to -ν). In Latin the vowel is shortened due to the general shortening of all long vowels in final syllables before all consonants except -s. In OCS the ending -ǫ is a regular reflex of a sequence of a back vowel followed by a nasal, following the loss of closed syllables in Common Slavic period.
4. instrumental singular: The oldest ending has been preserved in Vedic (senā "by an army"); the same ending is used in Classical Sanskrit, but on a stem extended with -y- by analogy to i-stems. In Balto-Slavic the reconstruction is uncertain: Lithuanian has acuted -à which could be formally matched with the Vedic ending, but dialects also show the ending -u (e.g. runku "by hand"), and certain adjectives have nasal -ą (e.g. mažą́ją "with the little"), which point to the ending *-ā́m. OCS has the ending -ojǫ taken from pronominal inflection (cf. тѫѭ (tǫjǫ, “with that”)). Nasal vowel also shows a pre-form *-ām, independently reconstructible also for Baltic. In Greek and Italic the instrumental case has not been preserved, but can be traced fossilized to certain adverbs, such as κρυφᾷ (kruphâi) / κρυφῆ (kruphê) "secretly", originally *"by means of secrecy" and Latin quā (“in what manner”) (possibly ablative though).
5. dative singular: PIE *-eh₂ey > post-PIE *-āy, yielding regularly Greek -ᾱι/-ηι and the rest of attested endings, except for Sanskrit which has it suffixed onto the -y- element taken from i-stems.
6. ablative singular: In PIE ablative singular of this inflection was the same as genitive, as can be seen in Sanskrit. Classical Latin ending -ā derives from Old Latin -ād, which was formed analogically to the ending -ōd of o-stems (see above).
7. genitive singular: PIE ending *-es has been regularly reflected in most of the branches. Latin -ae has final -i < *-ī by analogy to o-stems; the inherited ending has been preserved in fossilized phrases such as pater familiās (“head of household”) and adverbs such as aliās (“at another place/time”). In Sanskrit, the interfix element -y- is taken by analogy to i-stems.
8. locative singular: PIE ending *-eh₂i > post-PIE *-ay. In Lithuanian to the inherited form postoposition *en > e has been suffixed yielding attested -oje, while Sanskrit -yām has been taken from pronominal paradigm (cf. Sanskrit instrumental singular feminine तस्याम् (tasyām, “on that”)).
9. dual: As usual, reconstruction of dual case endings is very hard; in NAV dual characteristic is the ending -ih₁, shared by eh₂-stems with the nouns of neuter gender. PIE *eh₂ih₁ > post-PIE *ai, which regularly yields all forms except for Greek -ᾱ, which is by Mycenaean evidence shown to derive from earlier *-ᾰε. The rest of Greek inflectional endings (-αιν, αιιν) arose by analogy to o-stem forms. OCS final -u in GL agrees with final -os in Sanskrit, most probably pointing to PIE *-ows.
10. nominative/vocative plural PIE *-eh₂es > post-PIE *ās, regularly reflected in most of the languages. In Latin the ending -ae was taken by analogy from o-stems, while the inherited ending was preserved in Old Latin inscriptions (e.g. QVAS > quae, HASCAE > hae, RELIQVIAS > reliquiae). Greek also independently replaced the original ending by analogy to o-stems with -ᾰι. OCS -y continues the original ending for accusative plural (see below).
11. accusative plural: PIE *-eh₂n̥s > post-PIE *ān̥s, which has been preserved in dialectal Greek forms as -ανς, whence Attic -ᾱς is derivable. In most language the word-final cluster -ns has assimilated to -s. In OCS *ā merged with *ō, which was raised to *ū before a final -s, yielding *-ūns > -y.

Animate

1. nominative singular: The ending *-is comes from the zero grade of the morpheme and the nominative singular desinence *-s. It was preserved in Ancient Greek (e.g. πόλις (pólis)), in Latin turris, in Sanskrit अग्निस् (agnís), in PGmc *mundiz and Hittite ḫarkiš.
2. vocative singular: Sanskrit requires a diphthong, it could reflect either *-ey or *-oy.
3. accusative singular: The ending *-i-m comes from the zero grade and the accusative singular desinence *-m. It was preserved as Latin "-im", Sanskrit "-im". In both Greek and Hittite word-final PIE "*-m" was regularly replaced by /n/, yielding "-ῐν" and -𒅔 respectively.
4. instrumental singular: The ending *-i(y)eh₁ comes from the zero grade and the instrumental singular desinence *-éh₁.
5. dative singular: The ending "*-éyey" is securely reconstructible for PNIE. It was preserved intact in Sanskrit "-aye".
6. genitive/ablative singular: The o-grade in the ending *-oys is securely reconstructible on the basis of gothic "-ais".^[3] Anatolian however, points to "*-oyos"/ "*-eyos", which is probably a remodeling from "*-oys".^[4]
7. locative singular: PIE has the ending "*-ḗy", while Hittite has "-ai" that cannot come from a long "*ḗ", therefore the PIE ending must be "*-eyi". This would be composed from the full grade plus the locative desinence "*-i".^[4][1]
8. nominative/vocative plural: The ending "*-eyes" is attested in Latin "-ēs", Vedic "-ayas", and Hittite "-ēš".
9. accusative plural: "*-ims", composed by a zero grade in the root and the acc.pl ending "*-ms". Hittite in the other hand attests a full grade, but it's most likely analogical.

Neuter

Animate

Neuter

In Old English, the umlauted NOM/ACC forms are unexpected due to not being followed by a front vowel like **-iz etymologically.

Animate

Neuter

Note: Plurals for neuter n-stems, unlike descendant paradigms, were formed using the collective form *h₁néh₃mō, which is declined akin to the masculine n-stems (except that the nominative, accusative, and vocative forms were merged).

Note: Plurals were formed using the collective form *wédōr, which is also a heteroclitic stem. Furthermore, there is evidence that this declension paradigm was originally acrostatic, but descendants point to a proterokinetic paradigm by late PIE.

Masculine

Neuter

Note: The Masculine and Neuter paradigms are identical except in the Nominative/Accusative/Vocative forms. In addition, Lithuanian's neuter adjective is defective and uses the ancestral Nominative/Vocative/Accusative singular form for all of its cases.

Feminine

Note: Technically, Hittite has no feminine gender, as Feminine and Masculine nouns in other PIE languages are grouped together as "common" nouns.

1. nominative singular: For this case the ending is composed by the full grade of the morpheme and the nominative singular morpheme "*-s". The ending is *-onts the basis of Latin -ēns/āns < Proto-Italic *-onts/ānts as well as Gothic -onds < Proto-Germanic *-ondz.
2. accusative singular: The ending *-ónt-m̥ comes from the full grade and the accusative singular desinence *-m. It was preserved as Latin "-ontem", Sanskrit "-antam". Hittite generalized the thematic ending "*-om" at the expense of the original "-m̥", yeilding "-𒀭𒌅𒀭 " ("-antan").
3. instrumental singular: The ending *-n̥téh₁ comes from the zero grade and the instrumental singular desinence *-éh₁. It's supported by both Sanskrit and PGmc.
4. dative singular: The ending "*-n̥téy" is securely reconstructible for PIE. The accentuation is perfectly reflected in PGmc by Verner's law.
5. locative singular: The ending "*-énti" is reconstructible for PIE, Hittite "-anti" must be a recomposition of an earlier "*-ant͡s". The grade of the suffix vowel is uncertain.
6. nominative/vocative plural: For both PIE and PNIE the ending "*-ontes" is attested in Greek "-όντες", Vedic "-antas", and Hittite "-antiš".
7. accusative plural: The ending "*-n̥tḿ̥s" must be reconstructed, the zero grade in the suffix attested in Vedic.

• Alwin Kloekhorst (2008) Etymological Dictionary of the Hittite Inherited Lexicon, Brill Publishers
• Ranko Matasović (2008) Poredbenopovijesna gramatika hrvatskoga jezika, Zagreb: Matica hrvatska
• Andrew Sihler (1995) New Comparative Grammar of Greek and Latin, Oxford University Pres