Tuesday, 24 November 2009

"Path of Fire" and other dangerous paths: a translation of Bachchan's अग्नि पथ and some philological discussion (including Proto-Germanic *paþaz)

Harivanshrai "Bachchan" Shrivastav (November 27, 1907 – January 18, 2003) was a poet of Chhayavaad literary movement. One of his poems is अग्नि पथ ("Path of Fire"), which was used as the title for the award-winning film of the same name, starring his own son, Hindi film superstar Amitabh Bachchan.

अग्नि पथ
अग्नि पथ! अग्नि पथ! अग्नि पथ!

वृक्ष हों भले खड़े,
हों घने, हों बड़े,
एक पत्र-छाँह भी माँग मत, माँग मत, माँग मत!
अग्नि पथ! अग्नि पथ! अग्नि पथ!

तू न थकेगा कभी!
तू न थमेगा कभी!
तू न मुड़ेगा कभी!-कर शपथ! कर शपथ! कर शपथ!
अग्नि पथ! अग्नि पथ! अग्नि पथ!

यह महान दृश्‍य है-
चल रहा मनुष्‍य है
अश्रु-स्वेद-रक्‍त से लथपथ, लथपथ, लथपथ!

अग्नि पथ! अग्नि पथ! अग्नि पथ!

Path of Fire
The path of fire! The path of fire! The path of fire!

There may stand excellent trees,
they may be dense, they may be big.
But don't ask, don't ask, don't ask,
even for the shade of a single leaf!
The path of fire! The path of fire! The path of fire!

You will never tire!
You will never halt!
You will never turn!--
Give your word! Give your word! Give your word!
The path of fire! The path of fire! The path of fire!

This is a mighty sight--
Man is walking,
Drenched, drenched, drenched,
by tears, sweat, and blood!

The path of fire! The path of fire! The path of fire!

Some philological analysis:
Bachchan employs a high level of तत्सम (tatsama) vocabulary (words borrowed from Sanskrit), in preference to tadbhava (तद्भव) vocabulary (native Hindi/Urdu words inherited from Sanskrit) and borrowings from Persian and Arabic, in this poem. Thus:
  • Sanskrit अग्नि (agni) "fire" rather than the Hindi tadbhava (तद्भव) word आग (āg)
  • Sanskrit वृक्ष (vr̥kṣa) "tree" rather than something like Hindi पेड़ (peṛ)
  • Sanskrit मनुष्‍य (manuṣya) "man" rather than Hindi/Urdu आदमी (ādmī) [borrowing from Perso-Arabic]
  • Sanskrit पत्र (patra) "leaf" rather than Hindi पत्ता (pattā)
  • Sanskrit अश्रु (aśru) "tear (as in crying tears)" rather than Hindi आंसू (āṁsū) [Both Sansrkti aśru and its descendant Hindi āṁsū derive from PIE *dak̂ru "tear", which is also the source of English tear and Latin lacrima (> English lachrymose)]
  • Sanskrit रक्‍त (rakta) "blood" rather than Hindi/Urdu खून (khūn) [borrowing from Persian]
  • Sanskrit स्वेद (swed) "sweat" rather than Hindi/Urdu पसीना (pasīnā) [note that Sanskrit swed is obviously cognate with English sweat]
  • Sanskrit पथ (path) "path" rather than something like Hindi/Urdu रास्ता (rāstā) [borrowing from Persian]
Note that the last word is obviously cognate with English path, but this does not reflect a common inheritance. Rather English path is the result of a very early borrowing into Germanic (as Proto-Germanic *paþaz) from some Iranian language (cp. Avestan pɑntɑ, genitive pɑθɑ 'way', Old Persian pɑthi-), possibly Scythian, due to early contact between Germanic and Iranian peoples. Sanskrit path and the Iranian forms like Avestan panta/pɑθɑ (and thus the Proto-Germanic borrowed *paþaz) are all ultimately from PIE *pent (which is also the root underlying English find; see Watkins[1]:65).

We find this borrowed word path used early on in English (OE pæð), one of the few p-initial words in Old English (due ultimately to the rarity of the phoneme*b in PIE). One very evocative use of pæð in Old English is found in the following passage from the epic poem Beowulf, describing the hero's approach to the fen-lair of the troll-woman:
Oferēode þā æþelinga bearn
stēap stānhliðo stīge nearwe
enge ānpaðas uncūð gelād
neowle næssas nicorhūsa fela.
Beowulf ll.1408-11
The noble prince then traversed
the steep stone slopes, the narrow ways,
the tight single-file paths, the uncanny fords,
the precipitous headlands, the many nests of water-monsters.
These two poems are thus linked in their descriptions of difficult paths...and even employ the same word, path: due to the fact that Bachchan avoids the Iranian (Persian) borrowing rāstā "path", and because the Beowulf-poet opts for the Iranian borrowing path.

[1]Watkins, Calvert. 2000. The American Heritage dictionary of Indo-European roots. Boston: Houghton Mifflin, 2nd edn.


  1. Ah, yet another bit of evidence for the Eastern origin of Germanic. Don Ringe & Co. found that the best IE Stammbaum looks like this: Anatolian splits off first, Tocharian next, and Italo-Celtic next. Then comes Germanic, then Armeno-Greek, then Balto-Slavic, and lastly Indo-Iranian. (The evidence for Armeno-Greek is slender but quite solid.) Albanian is effectively irresolvable because the diagnostic morphology is gone and the lexicon is layer upon layer of borrowings from other IE languages, except that it can't be a sibling of any of the first three families.

    But that doesn't explain everything: the resulting tree is still inconsistent. It can be made consistent, however, by postulating three lexical borrowing interchanges (no way of knowing which way the borrowing went): one between Proto-Italic and Proto-Germanic, a later one between Proto-Germanic and Proto-Baltic, and finally a weak one between Proto-Greco-Armenian and Proto-Italic, which looks very odd geographically until we realize that Proto-Italic wasn't necessarily spoken in Italy at all.

    (Actually, I have no idea whether or not 'path' is already in the Ringe/Taylor database of IE lexical changes.)

    The other useful bit of evidence is the word Harvaða-fjollum in an old and mysterious bit of verse in the Saga of King Heidrek the Wise, which when you run it back through Grimm's Law turns out to mean "Carpathian fells" (with regular f > v in ON), shows that the Germanics spent time in South Central Europe.

  2. Oops, forgot to link Ringe & Co. The key papers are #10 (Ringe et al. 2002) for the basic tree and #4 (Nakhleh et al. 2005) for the three contact edges.

  3. Hi John. Thanks - I know Ringe et al.'s work on IE trees. In fact, on the basis of my interest in this work, I put in a (successful) request to invite Tandy Warnow as one of our keynote speakers (she's one of the co-authors in some of the Ringe et al. papers on IE trees) for the 2nd annual ILLS conference to be held here at UIll this summer.

  4. Thanks, very interesting stuff.....I seem to remember reading somewhere that Germanic path was of unknown origin....could you expand on that part a little? IN any case, a very interesting post, thanks.

  5. @theswain:
    Initial *p- in Germanic, if inherited, would have to derive from PIE *b-, a sound which is very rare in PIE, and in any case there are no other reasons to posit a PIE root *bend (and anyway there appears to have been a morphonological constraint in PIE against roots with two voiced,unaspirated consonants which would rule out *bend anyway).

    Another apparent Iranian loan into Germanic, again with initial p- is attested by Gothic páida "jerkin, coat", also OE herepād (Bwf. 2258).

  6. Thanks, that's very interesting stuff. I've read in recent years that some linguists think that Germanic was not a PIE language or was more heavily influenced by indigenous languages in areas they settled in part on account of the relative lack of initial p words. Could you drop me an email? n I'd like to talk to you about doing something in The Heroic Age and for an encyclopedia.

  7. I think one of the best bits of evidence for the glottalic theory is precisely this constraint against roots with two ejectives. It's hard to articulate *p'ent'- (the glottalic spelling of traditional *bend-), because you have to change the airstream direction twice in one syllable. Reinforcements are its greater typological normality and the rarity of p', which is the rarest ejective stop.

    invite Tandy Warnow as one of our keynote speakers

    Coolness! She's actually principal author on the papers that are CS rather than linguistics (me, I can handle both, despite having no degree in either).

  8. @John: Yes, I recall that glottalic-theory argument (though I still think it's problematic, given that then we have to assume that Germanic and Armenian are the more conservative languages w.r.t. to the change; and that the other language families (Italic, Greek, Indo-Iranian etc.) all coincidentally underwent the same change...)

    @theswain: The lack of p-initial words in Germanic can be attributed to the scarcity of *b in PIE. Though there are plenty of words in Germanic which don't have good PIE etymologies, so, yes, there is evidence for some sort of substratal influence in Germanic -- but Germanic is still clearly a PIE language. (Email sent, by the bye, to your gmail address).