The article is here: http://indiafacts.org/fallacies-proto-indo-european/, please leave comments there, not here.
10. The biggest fallacy and central to any discussion regarding the Protolanguage in IE studies is exposed by the presence of roots or more correctly dhātus ‘lexical seed-forms’ in Sanskrit. When all the paraphernalia of PIE reconstructions are laid aside the investigator finds that, in plain fact, only Sanskrit and Avestan (to a much lesser degree) have roots! The other IE languages have verbs and nouns etc. but not roots, as such, from which verbs and nouns etc. are derived. Even Sanskrit has many words that cannot be analysed or traced back to a dhātu (apart from borrowed words): e.g. kakud ‘peak’, nṛ/nara ‘man’, putra ‘child/son’, balakṣa ‘white’, śūdra ‘servile’ etc. But it has 2000 dhātus all told and about 700 fully active in the early language.
In his Dictionary, Walkins gives 5 roots ser, and of these he connects number 2 with S ̦√sṛ > sarati/sisarti ‘moves/flows/runs’ and then gets lost in the labyrinth of IE complexities. This |sṛ| is not found as an independent word noun or adjective, but is found in S as stem in sṛ-t ‘running’, sṛ-ta ‘having gone/passed’, sṛti ‘way’ etc. Then there are sara saraṇa, sarit, sāra, sārin etc. This is found also in a cognate form in Tocharian salate, in Gk hallomai and L salio, all meaning ‘leap/rush’, but only as verbs, not as roots and with very few derivatives. The most curious fact is that it’s derivative saras ‘eddy, whirl, wave, lake’ is in the name of the ancient river saras-vatī. This is cognate with Avestan haraxvaiti, also a river’s name; but there is no root nor other word connected with this harah in Iranian, so it stands alone!
The mainstream theory, that wants the common Indo-Iranian tongue and culture in Iran, says that the Indoaryans went to Saptasindhu and there gave their version of the name to a river to remind them of their former country. This of course is utter, wilful nonsense, because saras has a rich family of lexemes and a dhātu, but the Iranian haraḥ is a lonely orphan! So the movement must have been the other way round and the Iranians just lost dhātu and derivatives retaining only the name and memory of the river in Saptasindhu. (See§7-8.) Otherwise, it is impossible that the Indoaryans left Iran with only harah/saras and once in their new habitat started developing other lexemes and the dhātu √sṛ.
11. Of the 700 dhātus in the early Vedic texts, 200 are found in the root-form as nouns or adjectives and also stems for verbs. Thus, Vedic has √īś (m) ‘lord’ and verb īś-e ‘I reign’; but also derivatives īśa, īśin, īśvara etc. Similarly √ruc > ruc (f) ‘lustre’ and á-ru-ruc-at ‘one shone’ (in a past tense, called reduplicated aorist); but also derivatives ruk-ma, ruca-ka, rucin, rocana etc. Similarly √sad > sad (adj) ‘sitting’ and verb á-sad-at (aorist) ‘one sat’. 200 such dhātus with their families of derivatives (nouns and verbs etc) form a very rich inheritance – considering that no other IE language has anything. Tatiana Elizarenkova, the renowned Russian vedicist, put it like this: “the verb-root [=dhātu] is basic to both inflexion and derivation…it is irrelevant that for some root as such nouns are not attested” (1995: 50). Sanskrit has organic coherence.
The most telling aspect for the antiquity and significance of Sanskrit is precisely this organic coherence arising from roots generating verbs, nouns etc. This functions with the regular use of suffixes, verbal and nominal. I shall give only two examples, but the instances are hundreds.
S has the stems pad/pād- (weak/strong) ‘foot’ and √pad > vb padyate ‘befalls, falls’. Since the foot is the bodily part that in movement constantly rises and “falls” we see semantic as well as phonetic agreement. Gk has pous (Gen podos) and L pes (Gen pedis); Armenian, Hittite and Tocharian have similar cognates for ‘foot’. But none has a cognate verb like S √pad- ! Gm does have ge-fetan ‘to fall’ (Old English) and has cognates fôt/fuoz ‘foot’. Slavic also has pada/pasti ‘falls’ but no other nominal cognates. Lithuanian has the verb peduoti, but its padas is ‘sandal, shoe’ (not foot).
The IE cognates for “daughter” present a similar case. S duhitṛ for daughter is the √duh and the suffixes i-tṛ, as in pitṛ ‘father’, aritṛ ‘rower’, aśitṛ ‘eater’ etc. The verb is duḥ- > dogdhi ‘extracts, milks’ (hence duhitṛ = milkmaid!). Gk thugatēr, Gmc tohter, Sl dušti and Oscan (old Italic) futir, have no other plausible cognates in their total diction. Surprisingly, neither Latin nor Hittite have any cognations for IE daughter! The others have the noun but, not the verb.