nominals ending in 'an' and declension

funderburkjim commented 5 years ago

There are about 3000 entries in lexnorm-all2 ending in 'an', excluding cardinal numbers.

These are assigned models according to gender: m_an, f_an_I, and n_an.

The discussion below aims to be in agreement with published declensions in most cases. However, there are numerous 'irregularities' (described by Kale, Whitney, and Bucknell) that are not taken into account. I hope to deal with these irregularities, along with irregularities for other models, in some consistent way at a later time.

funderburkjim commented 5 years ago

endings for m_an model

Case	S	D	P
Nominative	A	AnO	AnaH
Accusative	Anam	AnO	naH
Instrumental	nA	aByAm	aBiH
Dative	ne	aByAm	aByaH
Ablative	naH	aByAm	aByaH
Genitive	naH	noH	nAm
Locative	ni*/ani	noH	asu
Vocative	an	AnO	AnaH

ni: This alternate is not present for nominals ending in 'Xvan' or 'Xman', where X is a consonant; such as brahman*.

funderburkjim commented 5 years ago

endings for n_an model

Case	S	D	P
Nominative	a	nI*/anI	Ani
Accusative	a	nI*/anI	Ani
Instrumental	nA	aByAm	aBiH
Dative	ne	aByAm	aByaH
Ablative	naH	aByAm	aByaH
Genitive	naH	noH	nAm
Locative	ni*/ani	noH	asu
Vocative	a/an	nI*/anI	Ani

ni, nI: This alternate is not present for nominals ending in 'Xvan' or 'Xman', where X is a consonant; such as karman. The description of this exception is most complete in Kale (p. 73, section 117); but I would like someone to confirm the absence of the 'nI' option in dual of cases 1,2,8 for neuters; e.g., to confirm that the Nominative dual neuter of karman is karmaRI, without the optional karmRI.

funderburkjim commented 5 years ago

formation of feminine

The approach taken to generate the feminine stem is:

If MW provides a feminine form (such as for yajvan f(arI), i.e. yajvarI), this is used
If the nominal ends in 'van', then that ending 'van' is changed to 'varI' (this is described in both Whitney and Kale). Note this rule could be used to explain the yajvarI example.
Otherwise, construct the declension of the word using the m_an model and replace the final 'A' of the instrumental singular with 'I'. For example Sarman (happy, prosperous) has masculine instrumental singular of SarmaRA, and SarmaRI is taken as the feminine stem.

The feminine stem so formed always ends in long vowel 'I', and is declined using the f_I model, like nadI.

Question on sIman

Deshpande (p. 194) mentions sIman (f. border) is to be declined like Atman (m.) (i.e. using the m_an model); doing this would imply, for instance, that the nominative singular feminine of sIman would be simA. However the steps described above (3rd step would be applicable) would imply the stem sImanI whose nominative singular would be the same sImanI; the current algorithm has sImanI. So, what is the true story? I haven't noticed sIman as an irregularity regarding formation of feminine. If it is irregular, are there others irregular in the same way?

funderburkjim commented 5 years ago

bases

The base used to combine with a particular ending has one of two forms; If the (last) pada (call it P) of the word is represented as Xan, then the two forms of the base are X or Xa; So for rAjan we have rAj, rAja as the two base forms; for karman we have karm, karma.

Note that any ending begins with one of the three letters: a, A or n.

When combining with an ending beginning with a or A, the X base is used.

When combining with an ending beginning with n:

Use the Xa base when X ends with a conjunct consonant ending in m or v.
- Example: Atman. Use Atma when combining with nA, resulting in AtmanA.
Otherwise, use the X base
- Example: laGiman (lightness). Use laGim when combining with nA, resulting in laGimnA.
- Example: takzan (carpenter). Use takz when combining with nA, resulting in takznA;
  - this is further modified by nR sandhi to final form takzRA
- Example: rAjan. Use rAj when combining with nA, resulting in rAjnA
  - This is further modified by using the palatal nasal following j, resulting in rAjYA.
  - The only cases (from MW) are compounds ending in 'rAjanand the wordmajjan. We decline this also as (in 3s)majj+nA->majjYA`
  - this special rule (rAj + n -> rAjY) appears to be a special sandhi.
    The only other MW word with a palatal preceding an is pARi-kUrcan. I've found no print authority to support the change of 'n' to palatal 'Y' after other palatals. The current algorithm declines kUrcan in 3s as kUrcYA.

funderburkjim commented 5 years ago

Example rAjan with m_an model

Case	S	D	P
Nominative	rAj + A = rAjA	rAj + AnO = rAjAnO	rAj + AnaH = rAjAnaH
Accusative	rAj + Anam = rAjAnam	rAj + AnO = rAjAnO	rAj + naH = rAjnaH -> rAjYaH
Instrumental	rAj + nA = rAjnA -> rAjYA	rAja + ByAm = rAjaByAm	rAja + BiH = rAjaBiH
Dative	rAj + ne = rAjne -> rAjYe	rAja + ByAm = rAjaByAm	rAja + ByaH = rAjaByaH
Ablative	rAj + naH = rAjnaH -> rAjYaH	rAja + ByAm = rAjaByAm	rAja + ByaH = rAjaByaH
Genitive	rAj + naH = rAjnaH -> rAjYaH	rAj + noH = rAjnoH -> rAjYoH	rAj + nAm = rAjnAm -> rAjYAm
Locative	rAj + ni/ani = rAjni/rAjani -> rAjYi/rAjani	rAj + noH = rAjnoH -> rAjYoH	rAja + su = rAjasu
Vocative	rAj + an = rAjan	rAj + AnO = rAjAnO	rAj + AnaH = rAjAnaH

funderburkjim commented 5 years ago

Example Atman with m_an model

Case	S	D	P
Nominative	Atm + A = AtmA	Atm + AnO = AtmAnO	Atm + AnaH = AtmAnaH
Accusative	Atm + Anam = AtmAnam	Atm + AnO = AtmAnO	Atma + naH = AtmanaH
Instrumental	Atma + nA = AtmanA	Atma + ByAm = AtmaByAm	Atma + BiH = AtmaBiH
Dative	Atma + ne = Atmane	Atma + ByAm = AtmaByAm	Atma + ByaH = AtmaByaH
Ablative	Atma + naH = AtmanaH	Atma + ByAm = AtmaByAm	Atma + ByaH = AtmaByaH
Genitive	Atma + naH = AtmanaH	Atma + noH = AtmanoH	Atma + nAm = AtmanAm
Locative	Atm + ni/ani = Atmni/Atmani -> Atmani	Atma + noH = AtmanoH	Atma + su = Atmasu
Vocative	Atm + an = Atman	Atm + AnO = AtmAnO	Atm + AnaH = AtmAnaH

funderburkjim commented 5 years ago

Example nAman with n_an model

Case	S	D	P
Nominative	nAm + a = nAma	nAm + nI/anI = nAmnI/nAmanI	nAm + Ani = nAmAni
Accusative	nAm + a = nAma	nAm + nI/anI = nAmnI/nAmanI	nAm + Ani = nAmAni
Instrumental	nAm + nA = nAmnA	nAma + ByAm = nAmaByAm	nAma + BiH = nAmaBiH
Dative	nAm + ne = nAmne	nAma + ByAm = nAmaByAm	nAma + ByaH = nAmaByaH
Ablative	nAm + naH = nAmnaH	nAma + ByAm = nAmaByAm	nAma + ByaH = nAmaByaH
Genitive	nAm + naH = nAmnaH	nAm + noH = nAmnoH	nAm + nAm = nAmnAm
Locative	nAm + ni/ani = nAmni/nAmani	nAm + noH = nAmnoH	nAma + su = nAmasu
Vocative	nAm + a/an = nAma/nAman	nAm + nI/anI = nAmnI/nAmanI	nAm + Ani = nAmAni

funderburkjim commented 5 years ago

Irregularity of han and its compounds

Although handling of many irregularities is being deferred, the current algorithm does provide a solution to the irregularities for han and its compounds. The algorithm is based on my interpretation of Kale's rule from page 75 (Section 111):

puzan, aryaman and nouns ending in han lengthen their a in the Nom. sing. only;
the n of han is changed to R after ha .

SEE @SergeA's interpretation of this rule in comment below.

Note 1: The example given by Kale is vftrahan (Indra, the killer of Vftra). I have interpreted the 'n -> R after ha' as applying to han as well as to all of its compounds. Note 2: Bucknell (p. 94) gives gohan as an example; and in that example he does NOT apply the n->R rule after 'ha': e.g., he shows 'gohanO' for Nom. dual, instead of 'gohaRO'.

Our declension algorithm has been revised; for gohan it agrees with Bucknell.

Further input from grammarians or examples from literature is needed on this 'n->R after ha' rule.

The other irregularity is that when the 'h' base is joined to an ending with first letter 'n', then the result is not 'hn' but rather 'Gn'.

funderburkjim commented 5 years ago

Example go-han with m_an model

Case	S	D	P
Nominative	goh + A = gohA	goh + AnO = gohAnO -> gohanO	goh + AnaH = gohAnaH -> gohanaH
Accusative	goh + Anam = gohAnam -> gohanam	goh + AnO = gohAnO -> gohanO	goh + naH = gohnaH -> goGnaH
Instrumental	goh + nA = gohnA -> goGnA	goha + ByAm = gohaByAm	goha + BiH = gohaBiH
Dative	goh + ne = gohne -> goGne	goha + ByAm = gohaByAm	goha + ByaH = gohaByaH
Ablative	goh + naH = gohnaH -> goGnaH	goha + ByAm = gohaByAm	goha + ByaH = gohaByaH
Genitive	goh + naH = gohnaH -> goGnaH	goh + noH = gohnoH -> goGnoH	goh + nAm = gohnAm -> goGnAm
Locative	goh + ni/ani = gohni/gohani -> goGni/gohani	goh + noH = gohnoH -> goGnoH	goha + su = gohasu
Vocative	goh + an = gohan	goh + AnO = gohAnO -> gohanO	goh + AnaH = gohAnaH -> gohanaH

Note: This comment has been changed in light of @SergeA comment below. This declension (now) agrees with that printed by Bucknell.

funderburkjim commented 5 years ago

example of vftra-han with m_an model

Case	S	D	P
Nominative	vftrah + A = vftrahA	vftrah + AnO = vftrahAnO -> vftrahaRO	vftrah + AnaH = vftrahAnaH -> vftrahaRaH
Accusative	vftrah + Anam = vftrahAnam -> vftrahaRam	vftrah + AnO = vftrahAnO -> vftrahaRO	vftrah + naH = vftrahnaH -> vftraGnaH
Instrumental	vftrah + nA = vftrahnA -> vftraGnA	vftraha + ByAm = vftrahaByAm	vftraha + BiH = vftrahaBiH
Dative	vftrah + ne = vftrahne -> vftraGne	vftraha + ByAm = vftrahaByAm	vftraha + ByaH = vftrahaByaH
Ablative	vftrah + naH = vftrahnaH -> vftraGnaH	vftraha + ByAm = vftrahaByAm	vftraha + ByaH = vftrahaByaH
Genitive	vftrah + naH = vftrahnaH -> vftraGnaH	vftrah + noH = vftrahnoH -> vftraGnoH	vftrah + nAm = vftrahnAm -> vftraGnAm
Locative	vftrah + ni/ani = vftrahni/vftrahani -> vftraGni/vftrahaRi	vftrah + noH = vftrahnoH -> vftraGnoH	vftraha + su = vftrahasu
Vocative	vftrah + an = vftrahan	vftrah + AnO = vftrahAnO -> vftrahaRO	vftrah + AnaH = vftrahAnaH -> vftrahaRaH

SergeA commented 5 years ago

the n of han is changed to R after ha .

This does not mean h to be trigger for n>ṇ change. This means the common rule n>ṇ (which is triggered by ṛ ṝ r ṣ) is allowed in -han compounds only after h, and not allowed if h is changed to gh.

funderburkjim commented 5 years ago

@SergeA Thank you for the interpretation of Kale's rule. Compounds of han now use my understanding of your interpretation -- See for example the revision to 'go-han' above.

Are there any edge cases which should be further checked?

Aside: What is the basis of your interpretation. Is it from familiarity with certain inflected forms appearing in literature, or extrapolated from certain Paninian sutras, or what?

SergeA commented 5 years ago

Paninian rule hanter at-pūrvasya | 8.4.22 | "of √han after /a/" goes under the main n>ṇ rule, so it requires to be preceded by a corresponding trigger. Conjugational forms are given as illustration: prahaṇyate, but praghnanti. In the case of ghn, n is not after a, so the change is not allowed. And it should be similar with the forms of vṛtra-han. But in go-han we have no trigger, so the n>ṇ change is not possible.

funderburkjim commented 5 years ago

other words ending in han

There are several other nominals in MW that

end in 'han'
are not currently declined as compounds of 'han' -- i.e., they are currently declined like rAjan.

compounds of ahan

These have their own differences from rAjan, but these differences have not yet been programmed. The words are:

ahan (day)
aGAhan (inauspicious day)
anahan (non-day)
ekAhan (single day)
kAhan (day of Brahmā )
mftAhan (day of death)
vyahan (done on separate days)
sAyAhan (close of day, eventide)

derived from other roots than han

mahan (greatness, from root mah)
snIhan (mucus, from snih)
snehan (a friend, also from snih)
snUhan (mucus, from snuh)
plihan (spleen, from plih ?)
plIhan (same as above)

tihan (sickness) -- from sickness, not sure if a han compound. Anybody know?

other

I think these are compounds from han, but not sure how to decline. Currrently declined like rAjan.

sadDan (Name of a deva-gandharva). MW says from han.
bfhadDan (as above)
amogGan (killing the stingy) .

gasyoun commented 5 years ago

but not sure how to decline.

Guess a list is needed to document all the questionable ones, otherwise they tend to get lost.

sanskrit-lexicon / MWinflect