Substrate language which contributed sarSapa to Indo Aryan
Therefore, no linguistic connection can be posited between this group of Munda words and Skt. sarṣapa ‘mustard seed’ (most probably a non-Indo-European substrate word).
Now let us examine what could be the substrate language.
Let us examine certain words in DEDR (Dravidian Etymological Dictionary):
Harvard-Kyoto convention is used; zh is used for retroflex approximant (frictionless continuant).
DEDR 1494 ‘kAzh’ (Tamil) blackness, blemish, defect; ‘kAzhakam’ blackness
DEDR 1278 ‘karu’ (Tamil) black
DEDR 1492 ‘kAzh’ (-pp-, -tt-) (Tamil) to be pungent, acrid; ‘kAzhppu’ pungency.
DEDR 1466 ‘kAr’ (-pp-, -tt-) (Tamil) to be pungent, acrid, hot to the taste, be very saltish or brackish; ‘kAram’ pungency.
DEDR1135 ‘kaTu’ (-pp-, -tt-) (Tamil) (noun) bitterness, pungency, poison, astringency; severe, cruel, harsh, extreme;
PD *-cc is another transitive/causative suffix (Krishnamurti, 2003, p 286, The Dravidian languages) Hence to be pungent can be ‘kAzh-cc-’.
If -p- the intransitive non-past additive is added we get ‘kAzh-cc-a-p-a’.
The retroflex approximant (frictionless continuant), in many Dravidian languages has merged with r.
Hence ‘kAr-cca-pa’ (also see DEDR 1466 ‘kAr’ to be pungent, above).
Rule 14 of Krishnamurti (2003) (Palatalization of velars) k > c occurs under different environments in different Dravidian languages. However such change is rare when ‘k’ is before vowel ‘a’ and is more difficult before long vowel ‘A’. There are examples involving long frontal vowels (Tamil word for yam ‘cEenai’ < ‘*kEen-’)
Hence ‘kAr-cca-pa’> ‘cAr-cca-pa’
There are many examples of Proto Dravidian –PP- to post Proto Dravidian –P- (here P = voiceless stop) in many languages (Krishnamurti, 2003, p177)
Therefore ‘cArccapa’ > cArcapa.
Now by Rule 13 of Krishnamurti (2003) (Affricate weakening, loss, irregular merger with dentals and velars), c > s.
Hence cArcapa > cArsapa > sArsapa (by rule 14)
‘Sarsapa’ by vowel shortening.
Compare this with Sanskrit word ‘sarSapa’, to get the non Indo European substrate language.
‘kAzh’ rather than ‘kAr’ (DEDR 1466 ‘kAr’ ) was used as the starting point for the derivation because another Dravidian word for mustard ‘kaTuku’ (Tamil) could be derived from ‘kAzh’.
zh > -D- in Telugu (which is equivalent to –T- in Tamil)
Hence ‘kAzh-u-ku’ > kATuku > kaTuku is possible. But more probably ‘kAzh’ > ‘kazhu’, because there is alteration between c1V1c2
> c1v1c2v2 in many stems (Krishnamurti, 2003, p 96).
And ‘kazu’ > kaTu from which ‘KaTuku’ is derived.
Again let us look at another construction:
According to Krishnamurti (2003, p154) Intervocalic -w- is derived from root-final *-p followed by formative suffixes in most cases. According to Krishnamurti (2003, p144) –p- is equivalent to –w- .
Therefore ‘cArccapa’ > ‘cArccava’ .
Initial c> s > zero happened in some of the Dravidian languages (eg. Tamil/Malayalam) (Krishnamurti, 2003)
Hence Arccava > Arccava
Kirshnamurti (2003, p 145) has also shown p > s > y in ancient Tamil.
Hence Accava > Ayyava could be a Tamil word for mustard.
Actually we find that another Tamil word for mustard as ‘aiyavi’ and also find ‘ayyavi’ in Malayalam (DEDR 921).
This deduction reconfirms that the Proto Dravidian word for mustard was ‘kAzhccapa’
Indo Aryan words are derived from it.
It is pertinent to note that there is no equivalent word like hariṣapa/harṣapa in Avestan/Persian language for the so called Sanskrit words सरिषपः sariṣapaḥ, सर्षपः sarṣapaḥ, as 'S' in Sanskrit becomes 'H' in Persian e.g sarasvati ↔ harahvati.